Probabilistic Anchor Assignment with IoU Prediction for Object Detection

By Kang Kim and Hee Seok Lee.

This is a PyTorch implementation of the paper Probabilistic Anchor Assignment with IoU Prediction for Object Detection (paper link), based on ATSS and maskrcnn-benchmark.

Note

Now the code supports PyTorch 1.6.

PAA is available at mmdetection. Many thanks to @jshilong for the great work!

Introduction

In object detection, determining which anchors to assign as positive or negative samples, known as anchor assignment, has been revealed as a core procedure that can significantly affect a model's performance. In this paper we propose a novel anchor assignment strategy that adaptively separates anchors into positive and negative samples for a ground truth bounding box according to the model's learning status such that it is able to reason about the separation in a probabilistic manner. To do so we first calculate the scores of anchors conditioned on the model and fit a probability distribution to these scores. The model is then trained with anchors separated into positive and negative samples according to their probabilities. Moreover, we investigate the gap between the training and testing objectives and propose to predict the Intersection-over-Unions of detected boxes as a measure of localization quality to reduce the discrepancy.

Installation

Please check INSTALL.md for installation instructions.

Inference

The inference command line on coco minival split:

python tools/test_net.py \
    --config-file configs/paa/paa_R_50_FPN_1x.yaml \
    MODEL.WEIGHT [/path/to/weight] \
    TEST.IMS_PER_BATCH 4

Please note that:

If your model's name is different, please replace PAA_R_50_FPN_1x.pth with your own.
If you enounter out-of-memory error, please try to reduce TEST.IMS_PER_BATCH to 1.
If you want to evaluate a different model, please change --config-file to its config file (in configs/paa) and MODEL.WEIGHT to its weights file.

Results on COCO

We provide the performance of the following trained models. All models are trained with the configuration same as ATSS.

Model	Multi-scale training	Multi-scale testing	AP (minival)	AP (test-dev)	Checkpoint
PAA_R_50_FPN_1x	No	No	40.4	-	-
PAA_R_50_FPN_1.5x	No	No	41.1	41.2	link
PAA_R_101_FPN_2x	Yes	No	44.6	44.9	link
PAA_dcnv2_R_101_FPN_2x	Yes	No	47.1	47.4	link
PAA_X_101_64x4d_FPN_2x	Yes	No	46.4	46.8	link
PAA_dcnv2_X_101_64x4d_FPN_2x	Yes	No	48.8	49.2	link
PAA_dcnv2_X_101_32x8d_FPN_2x	Yes	No	48.9	49.0	-
PAA_dcnv2_X_152_32x8d_FPN_2x	Yes	No	50.5	50.8	link
PAA_dcnv2_X_101_64x4d_FPN_2x	Yes	Yes	51.3	51.6	link
PAA_dcnv2_X_101_32x8d_FPN_2x	Yes	Yes	51.2	51.4	-
PAA_dcnv2_X_152_32x8d_FPN_2x	Yes	Yes	53.0	53.5	link

[1] 1x , 1.5x and 2x mean the model is trained for 90K, 135K and 180K iterations, respectively.
[2] All results are obtained with a single model.
[3] dcnv2 denotes deformable convolutional networks v2. Note that for ResNet based models, we apply deformable convolutions from stage c3 to c5 in backbones. For ResNeXt based models only stage c4 and c5 use deformable convolutions. All dcnv2 models use deformable convolutions in the last layer of detector towers.
[4] Please use TEST.BBOX_AUG.ENABLED True to enable multi-scale testing.

Results of Faster R-CNN

We also provide experimental results that apply PAA to Region Proposal Network of Faster R-CNN. Code is available at PAA_Faster-RCNN.

Model	AP (minival)	AP50	AP75	APs	APm	APl
Faster_R_50_FPN_1x	37.989	58.810	41.314	22.361	41.522	49.584
Faster_R_50_FPN_PAA_1x	39.292	60.019	42.567	22.650	43.170	51.875

Training

The following command line will train PAA_R_50_FPN_1x on 8 GPUs with Synchronous Stochastic Gradient Descent (SGD):

python -m torch.distributed.launch \
    --nproc_per_node=8 \
    --master_port=$((RANDOM + 10000)) \
    tools/train_net.py \
    --config-file configs/paa/paa_R_50_FPN_1x.yaml \
    DATALOADER.NUM_WORKERS 2 \
    OUTPUT_DIR training_dir/paa_R_50_FPN_1x

Please note that:

If you want to use fewer GPUs, please change --nproc_per_node to the number of GPUs. No other settings need to be changed. The total batch size does not depends on nproc_per_node. If you want to change the total batch size, please change SOLVER.IMS_PER_BATCH in configs/paa/paa_R_50_FPN_1x.yaml.
The models will be saved into OUTPUT_DIR.
If you want to train PAA with other backbones, please change --config-file.

Contributing to the project

Any pull requests or issues are welcome.

Citation

@inproceedings{paa-eccv2020,
  title={Probabilistic Anchor Assignment with IoU Prediction for Object Detection},
  author={Kim, Kang and Lee, Hee Seok},
  booktitle = {ECCV},
  year={2020}
}

kkhoot / PAA