GSA Image Recognition-AI
Classifying Images with Neural Network/AI
The GSA Image Recognition-AI program was developed for the automatic recognition and classification of image data. The program uses a neural network and artificial intelligence for classification.
The GSA Image Recognition-AI program was developed for automatic recognition and classification of objects in image data. The program uses a neural network to classify the input data, which can be configured and trained to meet the given requirements.
Features:
Features:
- — Resizing of the input images using different resampling methods
- — Image format conversion function (JPG, PNG, TIF, GIF)
- — Adjustable image filters (gaussian filter, edge selection, colour manipulations, ...)
- — Access to analysis function via included web server (HTML, JSON)
- — Batch processing function for testing optimal filter and neural network settings
- — Trained neural networks can be saved and opened with all parameters (input neurons, output neurons, weights, filters and filter sequence, ...)
- — The program can be universally configured for all different applications
- — Result output with user specifications (delimiters, output fields, ...)
Image classification with AI
The neural network attempts to assign the unknown images to an image group. The determined group and associated similarity in percent is shown as result.
Configuration dialog
This shows several options of the neural network such as image resolution, learning runs and interference factor.
Training room
The image groups are created in this interface where training images are assigned to their image groups and network training is started.
Control of image assignment
Assigned training images of an output group can be checked and corrected if necessary.
Correction window
Assigned training images of a defined group can be checked and, if necessary, corrected.
Batch processing
Network settings and image filters can be tested sequentially with their effects on the analysis results in a batch function.
Web-Service
The fully configured network can be operated via a web server which makes it easy to integrate the functions into external programs.
Conversion of image formats
All common image formats can be converted to other formats to minimizes the required training time.
Export function
The results of the analysis can also be exported to external files.