Some details
Development of a process to train CNN with synthetic images
Enigma Pattern was contracted to develop an unique method to generate and train convolutional neural network models basing on synthetic images. The entire process comprised:
- development and preparation of the environment basing on Caffe
- development of steering scripts for simulation of a natural environment using SynCity
- development
- modification of hyperparameters of the network in order to improve detection precision
- improvement of results by transformation of synthetic images
Results:
Repeatable process of building neural networks
based on synthetic images
See a separate presentation about results of this project.
Technologies:
Caffe, Unity3D