It will generate weights files for use by the dicebox service.Ī REST API that performs classification using the designated network structure and weights. The stand-alone training application for a selected network genotype. ( legacy) dicebox_multi_client.py - like the dicebox_client.py but classifies several portions of the screen.( legacy) dicebox_client.py - a python opencv thick client and supervised training program.dicebox_test_client.py - meant for bulk image classification and trained network validation.It is a simple html5/css/javascript client and supervised training program. webcam is the recommend client for most use cases.There are several client implementations included with dicebox. The Trainer uses the REST API for training data.The Web Client (and supervised trainer) stores data to an AWS EFS via the REST API.The trained neural network is accessed through a REST API.

In this case dicebox provides a means to evolve a network tailored to the particular problem. Sometimes unique or novel problems need to be solved and a neural network structure is unknown. It can be reconfigured for different image sizes and categories.ĭicebox is capable of being applied to a large variety of classification problems. The project includes several client implementations, and future enhancements will continue to expand the API capabilities.ĭicebox is a visual classification system. An image classification and training system built with SOA (Service-Oriented Architecture) in mind.

Dicebox givemeall