LINKS SIDE BAR

The Camera Store

Buy your Camera Equipment via Amazon

DSLR
Compact
Point and Shoot

To Return to this Page Click General Picture Recognition

Get Picture Recognition so you can always find that ONE Picture out of many.

..................

Version 5 Advanced Copyright 2008 to 2012 with On-line Help and E-mail Support

Please download the GPRSG SOFTWARE from here

E-Mail: GPRSG SOFTWARE for any further information or questions.

This software is unique and there is nothing else like it on the Internet. This software has been used by Port Authorities, Zoo's, Manufacturers and many Companies. It also has been used by Home Users all around the world. New updates can also be downloaded from the new free-update page for this current version 5. Computers hold tens of thousands of photos once stored usually difficult to find. This software uses Picture Recognition to find, tagging to store and quick search to retrieve.

The software is very easy to use but if you really wish to pin point an object within an image it also provides the end user with a host of methods to achieve just that within the professional version.

Please see the demonstration videos below they show its picture recognition ability, Quick tagging and a very easy to use retrieval system so that any image can quickly be found when needed.

Connecting GPRSG to GPRSG

Please see below for more information on all of these features.

If you have a castle and it looks very different say its from a different part of the world. Then searching against a sample will not help. To overcome this problem you can use searches made by other users around the world (this feature can be turned on or off). If on it collects information about the search you make and sends some analysis information to our server. This information can be used if someone else searches for the same type of object say a castle.


.

The dialog above is used to select the object in the sample image that you wish to search. It will find that same object in any other picture even if the size and shape has changed due to movement. For best results the objects need to have the same colour combination as selected. It is very complex to get good results and even more difficult if the colours are common colours. But with mode selection like "Object Colour Size Range" and with different mode selection for more common colours this advanced feature allows the User to search for a particular object within a picture. The results are very high for Object search.

.

This is the same dialog but with the picture analysed so that the helicopter can be identified. Usually the clearer the object to the original shape the better the result. However the wider shape modes can even identify the smallest of colour signatures. This is hopefully taking User defined object searching to the next level.


Home Version Quick View Table

  • Powerful General Picture Recognition
  • Normal Standard Object extraction method
  • Hobby Web-Cam Motion Detection
  • Hobby Time Laps Difference
  • File Tagging after Picture Recognition
  • Search Tagged Files found after Picture Recognition
  • Search the Internet via General Picture Recognition

Professional Version Quick View Table

  • Powerful General Picture Recognition
  • Advanced Powerful Object extraction method
  • Advanced File Tagging after Picture Recognition
  • Search Tagged Files found after Picture Recognition
  • Search the Internet Via General Picture Recognition
  • Picture Acquisition -Research and Professional Users
  • Select any object and it will attempt to isolate that object and
    compare it to the same object in any other picture using a special colour analysis method


Please download the GPRSG SOFTWARE from here

Please Download Demo Video of the General Picture Recognition Search (7 Minutes Long in HD Quality) using the General Picture Recognition Software. This is just one example of one of the many methods the software can use to find a particular set of pictures like fireworks.

Please Download Demo Video of the User Object Colour Search (5 Minutes Long in HD) using the General Picture Recognition Software. This allows the user to select any object within a picture so that the same object can be found in any other picture.

Please Download Demo Video of Organizer and Tagging (5 Minutes Long in HD) - this video shows how the software quickly organises and tags pictures. Once a picture is found using Picture Recognition the Organizer allows the User to Quickly see what other similar images are within the same folder and allows the User to Tag those images. Once an image is tagged it is easy to find again and even add or change the tagged information at a latter time.

Please Download Demo Video of Search for already Tagging Pictures (5 Minutes Long in HD) - this video shows how the software after Picture Recognition and Tagging Finds any Picture your looking for anywhere on the computer.

Please Download Demo Video of the Duplication software that comes as part of the GPRSG PACKAGE (2 Minutes Long) - this video shows how tags are transferred to all duplicates of an already tagged image. This is very useful if images are backed up to external drives. It gives the User the ability to search very large amounts of backed up images.

These demo video can be E-mailed to others, uploaded to sites like youtupe or any other web-site, used in any education or business establishment to demonstrate its ability. Different objects will give different results depending on colour within that object. But all in all it is a very good method of finding a particular user defined object in other pictures. To save the video just right Click on the link and Select (Save As) or just click to view.


  • Page Index Below
    Unique and Powerful General Picture Recognition
    General Explanation
    Main Purpose and Why the General Picture Recognition Software was Required
    Pre-word from GPRSG
    Software is Easy to Use but what it does is complex
    Home or Professional Version
    NEW Full Featured Tagging System
    User Defined Custom Colour Object Detection Analysis (Home Version Human Help)
    Connecting GPRSG to GPRSG
    Available Modes and Options
    Picture of Web Crawler with Picture Recognition
  • Home Version Hobby Features
    Time Laps Photography
    Motion Detection Photography
    Motion Detection Using Web-Cam
    Astronomy
    Research
  • Duplication Image Finder


Unique and Powerful General Picture Recognition

Mobile Devices like Iphones etc have become very popular, I have been asked if the software works with such devices. The solution is to download all your pictures from your device whether a phone, Iphone or any other device that connects to a PC. Run the General Picture Recognition Software or GPRSG Duplication Software that is also provided as part of the package that lets you remove all duplications and then upload back to the device. This is the only guaranteed way to assure that it will work given just about any device. However devices that create a drive will work and those devices can be searched directly.

Explaining the Concept

The concept is very simple if you have 10,000 photos and you do a search for say fireworks using the General Picture Recognition Software and you find 68 Fireworks the software will quickly let you tag those pictures. Each time you search for a group or class of pictures like fireworks, cars, trains, houses etc it will reduce that 10000 pictures down. Each time you do a search it will instead of search 10,000 pictures it will only search say 6000 pictures reducing down to zero. In the Pro version you can also select to exclude certain areas so it will not search those areas. Once tagging you can then quickly find just about any group of picture in minutes. If you wish to search a tagged image again for say a different object you can search already tagged images again using the general picture recognition engine. The software is very flexible. Once the image is tagged it can be found in minutes at any future point.

Please download the GPRSG SOFTWARE from here


General Explanation

Let your computer find your images using PICTURE RECOGNITION, unique software providing a unique solution for home and professional users. Let the GPRSG software do all the work of finding pictures.

Powerful Picture Recognition Software that attempts to Find any Picture Group like cars, trains, buses, trees, plants, aeroplanes, or just about any other group of photos you can think about. The GPRSG Pack also contains a very powerful Image ,video and non standard file Duplication Finder. The full pack is everything required to find images on the computer or Crawl the Internet looking for pictures with General Picture Recognition. The software gives you total control of your images from recognition through to advance group tagging of Pictures that have been recognised. The professional version is used by business to search large archives of images. While the home version is everything the home user needs to search for images on the home computer.

Picture Recognition (Plus) Full Tagging System (Equals) Images Found Quickly (Plus) Descriptive Information

Finds pictures that cannot be found on the computer. Search and find any picture with picture recognition. Have the best of both worlds, picture recognition to find the picture, tagging to quickly re-find the picture, and the traditional organiser to group pictures together.

Business Users

Due to the high commercial interest from Port Authorities, Museums, Zoo's, Manufacturers of well know products and lots of Private Companies that have inquired about the software. There is now a professional version of the software for version 5.


Main Purpose and Why the General Picture Recognition Software was Required

Most modern computers will hold around 20,000 to 60,000 thousand images usually put anywhere on the hard drive or a backup or network drive, if it took five seconds to view each image it would take about 5 second times 60,000 pictures =300,000 second divide by 60 second = 5,000 minutes divide by 60 will give 83 hours divide by 24 hours will give you 3.5 days rounded up. If you have say a million images and wished to find a picture taken a couple of years ago then it may take some time. With the General Picture Recognition Software it really does all the work of searching out those required pictures then lists the results so that tagging can take place.


Quality

This software is written with a high level of validation and quality in mind and is now in its 4th year of development. The General Picture Recognition Software itself is I think unique, and is a very Powerful full featured Picture Recognition Package.

There are two versions of the same GPRSG software the Home version and the Professional version. Both have many of the same features but the main difference is that the Pro Version has additional features that make it very useful for special types of searching that would not normally be required by the home user. For example the home version is there to search out the main object in the sample and once those objects are found it allows the end user to Tag those objects. Eventually all pictures once tagged will be easy to find in the future.

Home version only - has some interesting hobby type applications like motion detection in pictures and web-cam motion detection that has the ability to only detect a specific part of a picture. For example if you have trees that move you can simply rub them out or rub out say windows where lights go on or off etc. Any thing not rubbed out is detected via motion detection.

Professional version - motion detection, web-cam and any other hobby related features are not in the professional version. The professional version is designed more for pure picture recognition. It has some very complex features like the NEW "User Defined Custom Colour Object Detection Analysis" feature that is a very complex colour matching system for objects. This new feature is only available in the Professional version and replaces the less powerful "Human Help" within the home version (note that if you download the professional version the home version comes has part of the professional version as well.)

What is so impressive with "User Defined Custom Colour Object Detection Analysis" is that it can be used to find just about any object and as long as the selected colours are the same in sample and searched for pictures it will find that object. What is even more important is that size, position and shape is not important in colour analysis only that the colours are the same. So it can search out things like faces and is good enough in narrow mode to only find the same face excluding most other faces on a set of just faces or any other object you select. What makes it more impressive is that it distinguishes colour in such a way that it would seem to distingish between the same colour in different pictures. In other words if you have yellow in one picture and yellow in another picture but the objects are different it can usually understand that they are different yellows in terms of object. Sounds impossible but it provides a new way of isolating colour.

The end result is that you will just about every time find the same object in any other picture even if size, position are completely different. This NEW feature is only available in the Professional version.

Does this new feature have any down-sides, Yes it will not work with the same object if it has a different colour. Lighting may effect this type of search if the light changes the colour. However there are Narrow, Normal and Wide Modes just like in the Normal Picture Recognition Search.

Application - say you had several thousand pictures of Cars and you had a picture say of a Sample Car you wished to find, this mothod would look for that colour car within the several thousand cars and pull out any cars with the exact same colour features. Because it is analysing colour, in theory this function would even find pictures with say part of the car within it. The way it analyses colour gives the impression that it is distingishing between the same colour, of-course this is imposible.

Expectation - the GPRSG is a solid General Picture Recognition Software Product with many years of development. One of the main questions asked is can your product do what XXXXX say that they can do for example distinguish between different makes of Cars from any angle. The truth is No, finding a Car is difficult enough just to find a car unless the software was designed only to find different makes of Car and importantly is (dedicated only to find cars like face recognition software is dedicated only to find faces) it would be impossible to have software that could look at any object in the world and distinguish it from the same class of object and then do the same for any other object.

Currently the technology is usually good enough to find the same class of object for example a boat, car, train, house, aeroplane etc therefore most of the time it will be able to distinguish between say a house and an aeroplane but certainly could not distinguish between to different makes of aeroplane and then distinguish between different makes of cars and so on. Just to reiterate this is not to say that (dedicated) picture recognition cannot distinguish between cars, it possibly can, but it would not be able to distinguish between every other class of (object). This is what (General) Picture Recognition attempts to achieve in that it can search out ANY object it does not matter what that object is but at best can only do this down to a class level, car, house, boat it can distinguish usually between different classes but could not distinguish within its class i.e between two different makes of car or should I say I have never come across software that can. However this is not to say once a car is found that you cannot differentiate by tagging its make and model. One tagged you cn search for "Car", "Ford" "Escort" and it will pull out jut that car.


Software is Easy to Use but what it does is complex

The software is very easy to use just follow these three Steps:

  • Step 1 Get a Sample or Load in your own Sample Picture via file or cut and paste
  • Step 2 Choose a Location - Drive, Folder or Web-site to Search
  • Step 3 Run - Start the picture recognition software

That is it, however for more advanced user you will wish to control the search to get more customed results. Therefore you can use several advanced features to achieve the search.

Why is Picture Recognition so Complex

Here is a very small zoomed in section of a picture that the picture recognition software has to attempt to deal with below:

This section is mostly part of a castle wall zoomed into and the line going across the bottom of the picture is the top of a chair in front of the wall. This zoomed in picture is a very small part of the picture that you see below.

The first thing you notice when looking at both pictures is that the picture below seems to have less colour in it than the zoomed in picture this is because of the way pictures use a large combination of different colours pixel on pixel to give the effect of a certain colour. This makes it even more complex to determine what colour we are actually making a compare against. For example you may see a blue sky only to find that it has lots of other pixel colour that give the impression of a blue sky. You may have lots of white or pixels that are shaded red etc making up the blue sky. The arrangement of pixels can be just about anything so you may have several blue pixels then 1 shade of red pixel or two blue and 4 yellow then 5 blue. On a change of object you will find that several change over shades, even say from black to white you may have lots of gray and even a light blue etc in the mix.

Therefore when picture recognition software starts to find a match and it is doing it with some regularity then you should really be thinking WOW. Some pictures and objects are going to be easier to find than other pictures or objects. For example a less cluttered picture will always be easier to find if for example it has only one main object than a picture with one main object and several background objects say that run through the main object.

This software is easy to use for the beginner even if you do something wrong it will usually automatically take you to the right place for example if you try to run the software without having a sample it will ask you to choose a sample from the sample window.

On the other hand for those who are more familiar with the software you can use all of its advanced features to find most pictures you wish to search for and find, and gives the advanced user more options to home in on a particular image that they may wish to find.


Home or Professional Version

Home Version Main Window

This is a view of the main window within the Home Version the list below display the results and I have clicked on the first picture in the list below that was a Car. The outline picture is different because this was the last picture analysed before being stopped.

This is the organizer where images are copied or tagged or both. Once tagged instead of having to use picture recognition again to find the same picture you can use the built in Search Features, see below for more information to easily re-find any image tagged. The Professional version as more features.

Home Version

  • Main Feature, General Picture Recognition - on hard drive, backup drive and network drive.
  • New in Version 5 - full filename tagging system.
  • Mobile phones - that can connect to a laptop or PC via a drive can be searched.
  • Large View - view pictures that are being recognised, just click on a picture to large view it.
  • Manual Override - select any picture shown during picture recognition.
  • Web-Crawler with General Picture Recognition - searches out pictures on the Internet.
  • Web-Cam and DSLR or Digital Camera - picture motion detection.

Professional Business Version for Business Users

  • Both versions have the same general picture recognition engine, the pro version has improved initial search colour functions, but in the main they are the same.
  • Both versions have the New name tagging system.
  • Both version have all other features like the Web-Crawler with General Picture Recognition that searches out pictures directly from off the Internet.
  • The Business version does not come with Web-Cam or Motion Detection. The reason why Web-Cam and motion detection are not included in the professional version is that they are hobby type features, but if you like a versions features you can upgrade and run home and pro versions.
  • The professional business version is mainly for advanced needs within Picture Recognition.
  • The Business version comes with Picture Acquisition, that allows users within organisations to share pictures with other members in their organisation, even if you are in another office anywhere in the world, must be connected to the Internet for picture acquisition. Any user has the option of opting in or out. This is very different than just swapping pictures this is more about informing other users in the scheme what similar pictures to your sample are in different areas within your organisation (also a good feature for researchers).
  • Licences within the business version can extend to more computers than the limit within the home version of 3 computers. Please go to the download page for more information.
  • NEW - "User Defined Custom Colour Object Detection Analysis" very powerful object detection via colour analysis, advantages, size, position not really relevant.
  • In an office environment where time is very important the Professional version of the GPRSG Software has a built in (folder exclude or Ignore) area. The user can set-up the software to exclude certain system areas and custom areas to search. This will increase the speed at which new images are found. With tagging exclude as well (also in the home version) the software can very quickly analyse images that are in certain areas of the hard drive and not concern itself with image that are already known about.

Can I have both versions on one Computer

  • Yes if you have a laptop that you use at home and in work and you would like the hobby features from the home GPRSG and the Picture Acquisition for the office or research or some of the advanced features in the Pro version then you can have both versions running on the same laptop or PC. You can also install the full home package and upgrade to the Professional version later on using the On-Line Help - upgrade option this is a good option.

Researchers using the Professional Version -Enter in Company Ref field "research"

  • Instead of entering a company name or number into the unique Company Ref field any company or researcher can enter "research" into the Comp Ref field and if others outside of your company are using similar pictures to your sample and they join the scheme then they will have access to E-mails of any other users around the world who have jointed the scheme and are searching for the same or very similar pictures. This is a great way of building up contacts outside of your organisations, knowing that they have agreed to join the scheme and are willing to be contacted via E-mail.


NEW Full Featured Tagging System

  • Search images using picture recognition
  • Tag the results using the organiser
  • Search a Tag category for example Car and it will list all Cars that have been tagged in minutes not hours anywhere on the hard drive backup drive or network drive. Once the images have been tagged you can even search for those files using Windows 7 by using search commands like filename:"narrowboat*.*". Once tagged you can use windows itself to quickly find the tagged image or use the GPRSG's own tag search system with additional benefits of Category List. This searches all the tagged files within several minutes and lists each category. For example boat, car, train, hotels etc. Select a category and it will list all images of that category wherever they are on the computer.
  • Tagging allows the files to stay where they are but are easy to find
  • Using the Duplication Software you can also Auto Tag all Duplicate images

Now it is possible to find all the pictures on the computer using picture recognition then tag each picture so it can quickly be found again in the future. Individual and quick group tagging is possible making the software a very flexible tagging system.

This Version comes with a full featured tagging system

After picture recognition is complete you have the option of tagging groups of images found for example castles within the organiser or you can use the standard organiser functions or both. Once an image has been tagged you do not have to remember that much about the tag because the software will prompt you. You can do a quick search of either the main group see the most left hand list below or you can list the sub group see the second list box below. If for example you were looking for Goodwood festival of speed, you know it has something to do with cars so scroll to cars and see all sub-groups. Now click on a sub group "car, goodwood, general" and the Run menu will appear with the the search information in the three list boxes. Then click Start Tag Search and it will find images of Car that where at the Goodwood festival of speed. Because it is searching for a tagged images it will be able to do this in minutes.

That is how simple searching for an image is after you have tagged it using picture recognition. This software is ideal for people who store lots of images or who may have a hobby like photography or even a business. Backup your images to a external hard drive and the tagged information is retained. Move your image, copy your image and the tagged information is retained. Search for that image and if tagged it will find it in minutes, no indexing so if the index file ever becomes corrupt all that tagged information will NOT be lost. No need to scroll through thousands, all you ever have to do is think about the type of picture you wish to find and it will list the relevant group. The group information is extracted from the image file name therefore you know it will always find the correct image. Notice that the groups are in alphabetic order this makes finding that image a very easy task.

Viewing tagged images

Once the software as found an image using picture recognition and tagged that image so it can be found in minutes at any future point (see below) tagging not only makes it easy to search but easy to associate a picture to information. When scrolling through the pictures in the organiser it will auto display that tagged information relating to that picture. So if you have a picture of the red arrows, and you are looking at that picture it will automatically display tagged information like, "Red Arrows", "Swansea 2011". This is that all important descriptive information that tells you about that picture. Once tagged that information will always stay with that picture whether moved or copied to another folder.

After Picture Recognition

After picture recognition you will wish to do one or both things. Tag the image so that it is easily found in the future or group it together with other images that are the same.

What is tagging

Tagging simply allows you to associate a small amount of data to an image file. It makes searching for that image easier in the future and it describes what that image is about. So you have holiday snaps of beaches that you have been to around the world, the main category is beach the sub category is Brighton. You can search on beach that will list all beaches or Brighton that will list all photos of Brighton or Brighton Beach that will only list photos of Brighton beach. You do not even have to remember the category that a picture is listed under you can use the category search feature to list them for you.

More Options

Now you have two options using the organiser to organise your files by folder name that is very useful to group files together for example, put all pictures of castles in the same folder location. However sometimes you may wish to keep a file in its original location and still be able to quickly search out that image via the new tagging system.

Why Tag using the file name

This is the best option for tagging in my view. The history of image development means that different formats have been developed in different ways. There is no uniformed way to tag all types of images. The other big problem is copying from one format to another format will often mean that the tagged information is lost, this seems a well known complaint by others on the Internet. Tagging the file name means that the tagging process is independent of image type, this means you can copy or move a file and it will retain the tagged information. Even if windows renames the file it will leave the tagged information because it always adds to the end. The tagged information will look something like e.g. car-taggoodwood-sb1general-sb2anyothername.jpg you can even use windows to easily edit or remove the tag information if required. However adding and removing the information is easily done within the software because you can add or remove by groups.

When you should not use tagging

The tagging system is very safe in the sense that after Picture Recognition you can choose to tag any files anywhere on the disk, network drive etc. Some pictures will not be suitable for tagging but as the picture recognition system does not actually tag any images without the user agreeing first it is very safe. To tag a file after picture recognition you just select that file to tag. An example of a file that may not be suitable for tagging is an image file associated with a web-page, because tagging that file will change the file name of that image this may then require the web-page to point to the new file name.

How does the tagging system work

The tagging system is kept very simple, sometimes the best ideas are the simplest of ideas. Instead of creating indexes etc the tagging system will change the name of the file to the name of the main object in that picture. For example if you have a picture of a CASTLE and the file name is 3455443image.jpg by tagging that file using the picture recognition system it will change that file name to "castle-tag-generalsb1general-sb245544image.jpg" and leave it in the same location on disk. Tagging can use sub categories that provides information about the picture. For example you can sub tag and the file would look like this castle-tagedinburgh-sb1general-sb245544image.jpg. When viewing the file within the organiser it will extract the following "Castle" and "Edinburgh."

Searching for the tagged file

The software has a quick search method that will quickly search out the name of the tagged file. Enter "Castle" in the search box and it will list all the tagged files that start with castle. This will taken minutes not hours and will list all the tagged images of castles or any other tagged image. You can search via a sub category for example Edinburgh and it will list all images with Edinburgh in the sub category.

Ignore tagged files

You will automatically be able to ignore tagged images when running picture recognition. Therefore the more pictures you tag the less pictures will need to be searched each time you use the picture recognition software. It works like a sliding scale the more pictures tagged the less pictures to search through when searching using picture recognition. This option is defaulted On but can be turned off if you ever need to do a full search using the picture recognition side of the software.

How is this all accomplished

The system is very simple whenever you search for pictures using the picture recognition software it will list its search results. Continuing with the castle theme, if you ask the picture recognition software to search for "Castles" it will go off and search for images of castles. It will return all the castles that it has found. Say your computer holds 100,000 images it will search through and return approximately between 1 and 1000 images. The images it returns are moved to the organiser with a click of a button. You can then select the images to be tagged or organised by folder or both. Once tagged the pictures can be found again within several minutes not hours.

Once tagged you never have to tag that image again, however there are options to remove a tag and re-tag with another name at a later date. Now images can be left anywhere on disk but with the knowledge that you can access them in several minutes not hours or even days.

Exponential Tagging via the GPRSG Duplication Software

One an image is tagged once you can use the Duplication software to tag all other files that are exact duplications of the tagged image. This can be applied to backup images on a backup drive as well as normal duplications. This is automatically done by the duplication software if you decide you wish to tag all duplicate images with the same main object name.

Retrieve a tagged file anywhere on your disk

Just search for the category of a picture you wish to find and it will find it in minutes not hours. For example type in car and it will list all the image files with a car in them.

Cannot remember the category name then select Search Tag Category and it will list all categories of tagged image files on you hard drive in minutes not hours

Now you have picture recognition and now an easy to use tagging system, it does not matter how unorganized your images have become the GPRSG software does all the work of finding pictures and then tagging them so you never have to spend several hours searching for those pictures again. Searching via picture recognition then being able to quickly tag images for quick future searching.

Duplication Tagging means you can now automatically tag all duplicate image files

General Picture Recognition Software GPRSG now with a Full Tagging system and now get the home or business pack with the GPRSG Image Duplication finder that can now TAG all duplications completely automatically. This means that if an image has already been tagged using the new tagging system and there is any duplication anywhere for example a backup external hard drive it can tag that duplicate with the same tagged information as the original tagged file.

Now comes with a duplication finder that deletes images, video and other file duplications:

  • The Duplication finder is full featured
  • Very Easy to use


User Defined Custom Colour Object Detection Analysis (Home Version Human Help)

Now select an area of colour within a picture (object) and search for that colour (object) within other pictures. This is very useful if you know the colour you are looking for in both sample picture and the picture that is being searched. It will do an initial search and if it finds that colour within the picture it will analyse that picture.

There is also a new range function that attempts to look at the mass of the colour within a picture and search against that mass. If the area of colour has about the same mass as the target picture you can use the range function to search for that picture. This will narrow your search.

These methods allow the advanced user (person who has been using the software for a while) to customise a search. If the colour is unique not normally found in pictures it can actually be very fast for example searching for a gold colour is usually faster than searching for pictures with blue in them because there are a lot more pictures with blue in them than gold.

The colour search uses some very advanced techniques it is not just a colour on colour search.

User Defined Custom Colour Object Detection Analysis
(Advanced Features) Professional Version

Take a sample picture and identify ANY OBJECT within that picture and within the Professional version attempt to find that object in any other picture. The object can be ANY SIZE as long as the colours are the same. This type of search is done within the User Defined Custom Colour Object Detection Analysis Dialogue. The Professional version is much better suite to this type of search and can sometimes find a very small object given a larger sample. For those who need this level of functionality who already have the home version just go to the On-Line help Menu and upgrade to the pro version. You will then have the extra power of the professional versions of the software. The screen shot below is slightly older shot than the above but explains the basic concept well.


Available Modes and Options

The software comes with a large amount of flexibility with different modes for different situations:

Narrow Mode - this mode is very good if the sample is very close to the picture your searching, not exactly the same but close, for example the same castle from near enough the same location as the sample. Therefore if the main object is in the same location as the sample in the picture and they are the same object it will find it near 100% of the time. A.I is available in this mode.

Normal Mode - this gives you much more flexibility for example the castles can be taken from any location and even different castles all together it can find if they are similar in shape and colour but it is a castle. The accuracy levels will still be very good specially if it is for example the same castle just taken from different location. If however it is a completely different castle then the accuracy level will drop off a bit, but it still will find lot of different castles. A.I is available in this mode.

Wide Mode - this attempts to find a castle that is very different in size and colour. It will also use AI and past searches made via the Internet (can be switched on or off) to make compare against others who have searched for similar object. In this mode redundancy or false negatives increase but also the width of the search . Therefore although it will find many more pictures it will find many more castles than any other mode. This is the mode to use if you cannot find it in any other mode. However the software will find lots of pictures that are not castles but it will find castles that no other mode can find. This mode is going to be the slowest of all modes usually. A.I is available in this mode.

Landscape Mode - this will look at the colour composition in the whole picture and does not look for any particular object. It is a wide colour mode but if you have a sample of a beach it will be able to find other pictures of a beach or mountain range or forest. A.I is available in this mode.

Small Mode now called Fine Mode - this is used for collectors of say stamps etc., it does have several limitations the position and size need to be relative to some extent. When taking a photo of a stamp for example it must be on a very white background and same location if possible. This is the only mode that needs a clear white background.

Options Explained:

Using Slow mode the software can search each and every photo, however this would take a very long time. To speed up its search the software can be set to use one of the options below. The default option is Colour Quick Match 1 Auto - this option uses the colours in the sample picture to initially decide whether or not to further analyse a picture. It homes in on pictures with the same colour mix. At this point it is not interested in whether the shapes inside the picture are the same it is only interested in that picture if there are similar colour in both sample and target picture. Therefore if you have a sample of a castle it will use all the colour attributes of that picture to decide whether or not to analyse that picture further. Phase two of the analysis then starts to compare sample and target picture, with a more detailed colour and shape analysis to decide if it contains the same object and other attributes that make the picture similar.

Sometimes you may have a black and white photo that has little colour or no colour or is very monochrome. In this situation you would use the Black White Monochrome Match. It would then ignore pictures initially with very little or no colour and select pictures that are black and white or monochrome and then do a further analysis to attempt to identify shape. For those who are new to the software you really do not have to worry about the modes or options because the default mode and option are the best solution for most normal searches. Advanced users may wish to play with these options and modes to find a picture faster.

The Options are:

Slow - this is the slowest option and will analyse every single picture. Auto modes go into this mode to make a more intense check of a particular folder that it has found something in before returning to a faster mode.

Colour Quick Match 1 Auto - this option is the default search option and will flag pictures that have similar colours within them to the one that you are searching, it is very complex but usually very fast.

Colour Quick Match - this is not an auto mode but is even quicker than the default mode however it is possible to miss more initial matches with this mode but it is very fast usually.

Black White Monochrome Match - Sometimes you have a picture that has little or no visible colour in it. Using a colour match just will not find that picture. You should change to the Black White and Monochrome option. This will attempt to search out Black and White Photos - Use the FORMS option below if you are trying to find a scanned in form.

Full Match Mode - this is the fastest of all the modes and will only find an exact match but will go through at high speed. This mode can be varied to catch pictures that are not exact matches but a very close to exact match.

Object Center Color Match - this is matching the colors that are in the center of the picture it can be useful for example if you know that the picture you are looking for is similar in the center to the picture you are searching for color wise.

Forms - this is a very interesting mode if you are looking for a form that is mostly black and white it will search for those forms. It is used mostly to find a form that is scanned into the computer that is in a common picture format.

Pattern Only Auto - this is the only mode that really is concerned more about object and patterns to identify the picture not colour on its initial search. Therefore this mode should only be used if the object and pattern are a similar size and location. It is relatively slow option but can be useful now and again.

The software is set to its normal default mode and option when started. This will give the most common best results but after using the software for a while it becomes obvious which mode and option will give the best result for a certain search and sample. When using the web crawler it will default from normal to wide. This allows it to find more pictures when searching a web-site. The mode and option can be changed if for example you are looking for the exact same picture or a very close match.

Because you can run several instances of the software on faster computers for example 64 bit Windows 7 you can set up two or three instances using different modes and options to find the picture you require. For example run a very fast narrow mode and normal and just in case a much slower wide mode. With a strategy like this it will just about be impossible not to find that picture your looking to find. However most of the time it will find that picture with just one instance running.


Picture of Web Crawler with Picture Recognition

  • Crawl the Internet Searching for Pictures you require - Now Crawl the Internet searching for Pictures using the Picture Recognition Software with additional web crawler option. Now select to return not only pictures but web-pages that hold the key word or phrase. Now will return images and links to a word or phrase if found within any web-site.

    For example say you are looking for items like a piece of equipment you cannot find using a search engine and think that someone in a certain sector may have that item. Maybe your a coin or stamp collector you can take a picture of that coin or stamp and let the software crawl the web looking for that coin or stamp. You can select a web-site that is in that sector for example websites that are related to boats then search for a picture of boats the software will crawl the Internet looking for that picture using the picture recognition software that is directly connected to the crawler. If it finds a similar picture to your sample it will then display the web-site so you can look at other products you're attempting to find. If you are looking for say an Antiques this crawler function may be very useful. You can just leave the computer crawl away for hours and even days if you wish. You can also select different modes from narrow to wide you can also select it to find only very close matches. I have been astonished at the interesting things out on the web that you would have never have found using a standard search engine. It works by searching for a key word, once a key word is found it will start to analyse the pictures using picture recognition.

    This aspect of the software is very powerful because it can find pictures that are similar to your sample from any web-site in the world and its ability to crawl from one web-site to the next makes searching out pictures that are the same or similar to the sample picture in the database or one that you provide. If a similar picture is found you can then visit that web-site and view other similar pictures. The pictures from a web-site can be converted using the convert function into standard image files on your PC or Laptop. Once the conversion is made those files can be used in the Organiser allowing you to group common types of files found together.

    Force a key word search that will display links that not only contain the similar picture but links that have the same key word. This allows you to revisit a web-site with the product you are looking for even if there are no pictures or the pictures are with held server side. This is a very useful feature if you wish to find something on a large web-site like Argos that sever side protect their images. Its very helpful if you have several shopping web-sites, it will attempt to search them all and just list the relevant pages that you can quickly view in your own time. There will be a combination of pictures found by the picture recognition software and links found via the crawler.

    Phrase search for example if you are looking for make and model of a product and you only wish it to return pages with that exact phrase. This type of search will limit the amount of pictures returned but will only give pages with that key phrase in. Now the software can search out pictures, key words and now phrases.

    Cut and Paste any search engine page into the software and the software will extract those links. This is an easy way of starting to search through a group of web-sites that you have an interest in.

    There is built in protection so that this function is protected against most misuse. There are certain key words that the software will not allow. If those words are used the function will not work and you will be prompted that an inappropriate word has been used. Certain words maybe exempt that you may not consider should be exempt but those words are exempt because if used may result in it catching pictures that is should not catch and maybe of an inappropriate nature. Although there is built in protection if this function is used by children they should always be supervised by an adult. The crawler part of this software CANNOT be used for pornography it will NOT search inappropriate key words.

  • Part Site Crawl - Crawl through a particular web-site from a particular location goto a section and then use pre-load and start crawling from any location inside a very large web-site like e-bay. Use Picture recognition to find the pictures and location you require.


Home Version Hobby Features

  • Time Laps Photography - use your camera to take pictures of a certain area and then download the pictures from your Camera and create time laps photography picture sets.
  • Motion Detection Photography - set-up your Camera on a tripod and using the timer on your Camera or Intervalometer for DSLR. This is a powerful feature that will allow you to record movement thoughout the day in a certain area. If the camera takes say 5,000 pictures throughout the day you can use the software to extract pictures with actual motion in them. Pictures without any motion are ignored. From Bird Life to Deer or if you are say a researcher the amount of people or cars or boats moving through a certain area during a day or using a certain bridge etc. No expensive hardware required just use your normal digital camera. Motion detection via a good quality DSLR with give you perfect quality picture at a huge distance with the correct lens.
  • Motion Detection Using Web-Cam - this is I guess the cheapest option. The web-cam aspect of this software is limited in that it's only been tested with a (PC-World 1.3 Meg PC-LINE WEB CAM I CAN CONFIRM THIS WEB-CAM WORKS WITH THIS SOFTWARE, Microsoft LiveCam VX-700 WEB CAM TESTED AND WORKS WITH THE SOFTWARE. The microsoft web cam VX-700 was more expensive but has auto focus this means if the web-cam changes distance it does not need manually to be adjusted.). Once you have a web-cam connected the software itself that actually detects the motion is very powerful indeed. It's default mode is (Automatic) and can detect anything from a very small spider to a ship in the sea. It locks on to any area, you can move the web-cam around and once positioned the software will lock onto that new position. The biggest downside is that your limited to the picture quality of a web-cam against a camera. The up-side is that the Web-Cam option runs in real time via the software. When running in real time it can highlight motion in any area by displaying that motion in yellow. This is very interesting because you actually see the movement before you would using the human eye or even miss completely using the human eye. I only noticed this when testing the software myself. I found that the software saw things that I had missed when looking at the same area. I also have some amazing photos of a spider who decided to walk right up to the web-cam.
  • Astronomy - have been testing the software with a new telescope, the general idea is take a web-cam with auto focus connect it to a telescope to give a live feed of the moon, stars etc, then either use the motion detection or take snap shots. I have already tested it with binoculars and have found it easier to just point the binoculars and view the screen. There is an island out to sea I have already tested with binoculars but way to far away to sea any bird life. Just purchased a telescope to do some further testing will show results when completed.

    Research

    Motion detection part of the General Picture Recognition Software is a valued tool for researchers who wish to research the movement of just about anything day or night. The moon detection is full featured with auto light sensing and auto position. Even if the camera position is changed or the light changes the software will adjust and continue to operate.

    This is a large screen view within the software so that the user can see the web-picture in a larger view. You can also select yellow find that will highlight in yellow any motion in the picture and highlight that change in yellow. This is a large view of the motion detection screen on the right.

    This is the same man from the larger picture here he is moving away at a distance and being tracked by the motion detection software. Motorbike into car park motorbike leaving car park. Because you can set the mass of whatever you wish to track it can ignore smaller objects moving in the same picture. Also you can set it to only detect motion in a certain part of the picture.

  • Difficult to get to locations - most web-cams are very small, you can use them through small holes and see what is going on behind that location. All you need is a web-cam and a portable laptop and you can look at locations you would find very difficult to view and with this software even more importantly analyse any movement in those difficult to reach locations.
  • Security and motion detection - Although the web-cam software is not a security system that can cost a large amount of money you can in fact buy say the PC-line web cam for a couple of pounds and have a very simple security system that will allow you to monitor what is going on outside your home and capture any motion in a certain area. All motion is captured using JPG still images.
  • Larger screen view - of that location taken by the web-cam. You can have a larger screen view of the web-cam picture in real time to view in detail what you are seeing.
  • Motion Detection - Most people today have a digital camera or web-cam with this software you can now turn your camera or web-cam into lots of different things from a simple security cam, motion detection for things like Bird Watching using either a camera or a web-cam or if you are researching, say how many people cross a bridge etc. this software can be used in any area of motion detection.
  • Picture Recognition and Motion Detection - Feed any still picture into the picture recognition engine. Motion detection is very different from picture recognition. With motion detection it will pick up any motion. It is not trying to attempt picture recognition. However this is when it becomes really interesting. Say you took photos of a certain location say 1 year ago and you were at that location again taking more motion detection images. You can move a picture into the picture recognition search engine and search your drive for similar pictures. It will then go off and attempt to find pictures taken previously of that location so you can compare then with now.

Finding those Impossible to find Drawings, Forms or Flowcharts

If you keep Drawings or Flowcharts in normal picture file formats then the picture recognition software should be able to find them given a good sample. One very good use of the software is its ability to find very close matching documents. If you scan in a blank form and you have a form on the computer already filled in the software will usually be able to find that form or flowchart.

Hobbies like Astronomy or Bird-Watching

Connect your computer to an Auto Focus USB Connected Web-Cam, using either binoculars or telescope and zoom in via your laptop to objects a very long way away. The island in the distance is taken with a normal web-cam. The same island can be seen through binoculars. The higher the resolution of the web-cam the larger the picture and the more detailed the picture.

This is the web-cam again connected to the binoculars via the computer looking at tree branches close up. No birds found on this occasion but it gives the basic idea. The GPRSG software can motion detect on still days or snap shoot on windy days. Also pictures can be stored organised and new photos can be checked against photos taken in the past by moving them to the picture recognition module.


Duplication Image Finder

GPRSG image DUPLICATION SOFTWARE when you download the GPRSG Software. In this version the duplication software attempts to analyse images that are used a lot and attempts to protect you from deleting those images. If there are several images that are the same it will suggest less used duplications that should be considered for deletion. Also the software will attempt to block deletions from certain folders where it thinks it is unwise to delete those images. However you can turn this type of protection off and override it. When you download the full version you will be downloading the new improved version 5 image duplication software with the full GPRSG Package. It is possible to order the GPRSG Duplication Finder stand alone software for only £5.00. This is one of the most safe and powerful Duplication finders on the Internet.

Please go to the download page from here

Legal Notice: this web-site content and software is governed by United Kingdom Law of England and Wales. Terms and Conditions of use are available on request.

Demonstrations

If your a Small, Medium or Large Size Company within the geographical area of England or Wales (UK) we will demonstrate the GPRSG Software to your organization with a comprehensive demonstration of how the General Picture Recognition Software (Professional version) could benefit your company. Get a first hand view of how it works and see what it can actually achieve.

Please note there is an £80.00 demonstration charge plus travel expenses this will be calculated and sent via E-mail with available dates.

To arrange a demonstration E-mail: gprsg@hotmail.co.uk and we will get back to you in 5 working days.

Search and Tagging Service

Your Company, Organition or Small Business may have several thousand pictures that need to be grouped and tagged but you do not have the manpower or time.

England and Wales (Uk) only, we can come to your company develop a stategy and setup our software and search via picture reognition, tag most of your images and train staff how to easily retrieve those picture and use the software.

We only charge for the cost of the software (pro version), cost for example tavel etc and only £6.00 per hour labour (minimum) hours of 8.

Send us your location and we will cost a dailly rate and send it to you via E-mail before you make a decision.

If you require this service please contact us via E-mail: gprsg@hotmail.co.uk

Version 5 - Download the GPRSG Picture Recognition Software (Package).

We believe this is the most powerful General Picture Recognition software on the Internet.

PICTURE RECOGNITION now with a full Tagging System and File Organizer.

Home and Professional versions available to meet every need.

This pack includes the GPRSG Duplication Finder Software.

Free Upgrade Support for all version 3 and 4 users . Upgrade to version 5 free.

For version 5 users please download the latest updated version 5 minor releases from here.

Download the GPRSG Duplication Finder full featured) Advanced Version v5.0. If you only need software to find duplications (exactly the same image but somewhere else on the machine so that they can be deleted then this is the software required.

Built in File Monitoring that helps protect against deleting often used duplicates and protects from deleting images in protected areas like certain system folders. It also comes with a large amount of other features including near match.

IF YOU LIKE THE DUPLICATION FINDER IT IS EASY TO UPGRADE TO THE FULL PACKAGE. JUST CLICK UPGRADE ON THE DUPLICATION FINDER SOFTWARE.

Software Guarantee - Excluding PayPal expenses there will be a refund given if the home version for any reason does not live up to expectation within 7 days after activation of the software..

For the trouble of downloading the software that does not meet your expectation any software downloaded can be kept.

The Pro version is not protected in the same way therefore if you wish to see how suitable the software is for your needs please download the home version then via the home version upgrade to the Pro version.

Conditions

There are two conditions that is feedback on exactly what the problem was will be required. The reasons will be used to improve future development.

The software is general picture recognition not dedicated therefore it should not be expected to meet any dedicated application requirement.

Apart from these condition a refund will be guaranteed if the software does not live up to expectation within 7 days.

Any Questions Please E-Mail: gprsg@hotmail.co.uk

Feedback and New Features

Please E-mail if you would like to see any new features in future versions of the software.

Please continue also to send feedback on any aspects of the software you really like.

gprsg@hotmail.co.uk

Improved Redundancy or false negatives:

The most difficult aspect of picture recognition is not actually finding a match. The hard part is actually reducing the amount of redundancy or false negatives.

Version 4 has improved on version 3 in reducing the overall redundancy. On most searches it is down to less than 10%. With some samples even down to 0.2%.

Accuracy Levels

Apart from the estimates below in very simple terms the more complex the picture the more difficult it is usually to make sense of that picture.

The main question asked is how accurate is the software. If you have the same picture it will always find a duplication every time 100%.

If you have an object that has a fairly clear background and is the same size and shape and colour as the sample again it is near 100% if not 100%

If you have the same main object but with very different backgrounds then it will start to reduce say about 80%.

If you have the same main object different colour different size and different shape but same object it will start to fall away between 60 to 40% approximately. However if the object as been found in the past or picture

If you add a lot of background differences and clutter then possibly down to about 20%

Because the software is deigned to look for any object and any picture you can imagine overall the accuracy levels should be considered very high indeed.

But even these percentages are just averages because even different samples of the same object may give different results.

Changing from normal to say wide mode will increase redundancy but also increase percentage in terms of finding the picture required.

Also autonomous and real thinking impact on accuracy levels.

Currently I am unaware of any other software that attempts to achieve the same goal of general picture recognition with any higher accuracy levels.

What is considered impossible:

The theory that makes it possible.

"Relative in object theory" is my theory and my theory alone.

To comply with the theory:

  • All objects must be relative to each other
  • An object in one image can be matched and relative to any other object in any other picture.
  • That object must be relative to sub internal object and external object.
  • Objects that are not relative are considered new objects in their own right.
  • Objects once learned must be matched with the same object unlearned
  • Objects that are identified and matched are used to explain the picture.
  • Each object identified and know creates a word. Each word creates a sentence that explains the picture.
How far away is this theory from reality.:
  • Currently my software is dealing with a single object within a picture called the main object. It has lots of other parameters but in the whole it is only attempting to find a single object in each picture. For example a castle or car or train etc.
  • It does not attempt to make sense of every plant, tree, window or pain of glass or even type of cloud etc. Eventually to make true sense of all pictures we will need to make sense of each and every object.
  • Computing Power required to achieve this will be huge.
  • Internet structure - the Internet is the ideal place to build input and gather information to make sense of pictures. It will eventually be the main source for achieving this because a picture is just a huge amount of information that requires sense to be made from it.
  • Will system ever be able to achieve this, YES, eventually it will be possible to view a picture and a detailed - computer driven explanation will be made of that picture including, location, and a full explanation of what that picture is all about without any human involvement.
  • DNA, is the nearest thing I can think of to describe how pictures will be understood in the future. Each aspect of each and every picture ever created will have its own DNA associated to it. This is basically my theory of "Relative in object theory." The DNA needs to be learnt and then applied relative.
  • General Picture Recognition will eventually become the norm with accuracy levels being about 99.999%
But why:
  • Take any photo and one day you will have instant feedback of location, what that photo is about, who, description of every aspect of that photo. For example if the photo includes a boat one day the software will be able to tell you what type of boat it is who it is registered to what class of boat it is and so on. The boat may just be in the background of the picture but one day all relative information relating to that picture will become available to the end user.
  • Knowledge associated with a picture, it could be huge if you can break up each individual object within a picture and relate it to other known objects in other pictures that themselves are related and validated to databases of information with the ability to link any unknown object with relative information then a picture will one day be more important than a thousand words.
Question and Answers:

Question

Why when it finds the picture and I have tagged that picture next time it cannot find that picture using picture recognition.

Answer

Once a pictures been tagged the picture recognition engine will ignore it during a new search. However if you wish for some reason to search that picture again then click the Tag checkbox off. It will then search all pictures even ones that have been tagged

Question

It found other pictures but not a picture that related to the object I was searching for and I was using normal mode.

Answer

The next thing to do is try wide mode. If it does not find it using wide mode then use a different sample. But also it maybe down to the fact that there is not another picture with that same object or there is only one or two pictures and they are different from the sample's main object.

Some pictures if they are very different from the sample may just not be found.

Question

This software not only seems to find objects that are near to the sample but it also seems to find the same object that is black and white when the sample is colour. How does it do this, that's impressive.

Answer

There is a large amount of technologies that allow it to search out picture. It uses something called AI that helps and also it does not just use colour matching but also shape matching technology.

Question

Why is this software not in the shops it's great.

Answer

This started off as a hobby type development but maybe one day it will be in the shops.

myspace visitors