This is a PR to support RetinaNet, a state-of-the-art one-stage object detector. The model (ResNet 50 with 800 pixels) achieved AP of 35.3 instead of 35.7 reported on the paper.
you can download my PyTorch weights here (AP 35.3)
this code allows you also to load Caffe2's weights of RetinaNet. You can download the weights here (AP 35.7)
This is a PR to support RetinaNet, a state-of-the-art one-stage object detector. The model (ResNet 50 with 800 pixels) achieved AP of 35.3 instead of 35.7 reported on the paper.
Details of results are shown below:
RetinaNet
retinanet-R-50-FPN_1x
Testing command with PyTorch weights:
Testing command with Detectron weights: