Papers › Libra R-CNN: Towards Balanced Learning for Object Detection

Libra R-CNN: Towards Balanced Learning for Object Detection

4 Apr 2019CVPR 2019 6arXiv:1904.02701archive 2025-07-28

Jiangmiao Pang, Kai Chen, Jianping Shi, Huajun Feng, Wanli Ouyang, Dahua Lin

Compared with model architectures, the training process, which is also crucial to the success of detectors, has received relatively less attention in object detection. In this work, we carefully revisit the standard training practice of detectors, and find that the detection performance is often limited by the imbalance during the training process, which generally consists in three levels - sample level, feature level, and objective level. To mitigate the adverse effects caused thereby, we propose Libra R-CNN, a simple but effective framework towards balanced learning for object detection. It integrates three novel components: IoU-balanced sampling, balanced feature pyramid, and balanced L1 loss, respectively for reducing the imbalance at sample, feature, and objective level. Benefitted from the overall balanced design, Libra R-CNN significantly improves the detection performance. Without bells and whistles, it achieves 2.5 points and 2.0 points higher Average Precision (AP) than FPN Faster R-CNN and RetinaNet respectively on MSCOCO.

PaperPDFConference PDFCode

In Syntology Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

Code

open-mmlab/mmdetection officialmentioned in paperpytorchApache-2.0 report
OceanPang/Libra_R-CNN mentioned in papermentioned on GitHubpytorchApache-2.0 report
CVUsers/Smart-Retail-By-Efficientdet mentioned on GitHubpytorch report
hualuluu/--every-day-paper-- mentioned on GitHub report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Object Detectionobject-detection

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Object Detection COCO minival Libra R-CNN (ResNet-50 FPN) AP50 59.3 #195 of 220 Archive leaderboard report
Object Detection COCO minival Libra R-CNN (ResNet-50 FPN) AP75 42.0 #195 of 220 Archive leaderboard report
Object Detection COCO minival Libra R-CNN (ResNet-50 FPN) APL 50.5 #195 of 220 Archive leaderboard report
Object Detection COCO minival Libra R-CNN (ResNet-50 FPN) APM 42.1 #195 of 220 Archive leaderboard report
Object Detection COCO minival Libra R-CNN (ResNet-50 FPN) APS 22.9 #195 of 220 Archive leaderboard report
Object Detection COCO minival Libra R-CNN (ResNet-50 FPN) box AP 38.5 #195 of 220 Archive leaderboard report
Object Detection COCO test-dev Libra R-CNN (ResNeXt-101-FPN) AP50 64 #164 of 225 Archive leaderboard report
Object Detection COCO test-dev Libra R-CNN (ResNeXt-101-FPN) AP75 47 #164 of 225 Archive leaderboard report
Object Detection COCO test-dev Libra R-CNN (ResNeXt-101-FPN) APL 54.6 #164 of 225 Archive leaderboard report
Object Detection COCO test-dev Libra R-CNN (ResNeXt-101-FPN) APM 45.6 #164 of 225 Archive leaderboard report
Object Detection COCO test-dev Libra R-CNN (ResNeXt-101-FPN) APS 25.3 #164 of 225 Archive leaderboard report
Object Detection COCO test-dev Libra R-CNN (ResNeXt-101-FPN) box mAP 43.0 #164 of 225 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Methods

Introduced by this paper: Libra R-CNN

1x1 ConvolutionAverage PoolingBalanced Feature PyramidBalanced L1 LossBatch NormalizationBottleneck Residual BlockConvolutionEmbedded Gaussian AffinityFPNFocal LossGlobal Average PoolingGrouped ConvolutionIoU-Balanced SamplingKaiming InitializationLibra R-CNNMax PoolingNon-Local BlockNon-Local OperationReLUResNeXtResNeXt BlockResidual BlockResidual ConnectionRetinaNetStep Decay

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections