The code is for "iffDetector: Inference-aware Feature Filtering for Object Detection". This repository provides iff detector heads for YOLOv2, FoveaBox, and FreeAnchor. The corresponding config files are in the configs folder.
Detector | Backbone | Dataset | mAP | Model |
---|---|---|---|---|
YOLOv2 | Darknet-19 | VOC07 | 60.4 | Link |
Detector | Backbone | Multi-scalet testing |
AP(test-dev) | Model |
---|---|---|---|---|
FoveaBox | ResNet-50 | N | 37.6 | Link |
FreeAnchor | ResNeXt-101 | Y | 47.9 | Link |
Please refer to the following Code Link for corresponding requirements and usage: YOLOv2, FoveaBox, FreeAnchor. Note that according to the Linear Scaling Rule, you need to set the learning rate proportional to the batch size if you use different GPUs or images per GPU.