JdeRobot's DetectionSuite is primarily used for fair comparision of different Deep learning models on any framework or dataset preferred.
This project proposal aims to expand DetectionSuite’s current functionality to support more deep learning Frameworks such as Tensorflow, keras and Caffe and more datasets such as Imagenet, COCO, etc. It also aims to add Segmentation support to this tool along with various evaluation metrics in order to compare different detection models. For easy installation, this project will also improve current documentation and create debian packages.
Link to Proposal