Hybrid Task Cascade for Instance Segmentation

CVPR 2019 Kai ChenJiangmiao PangJiaqi WangYu XiongXiaoxiao LiShuyang SunWansen FengZiwei LiuJianping ShiWanli OuyangChen Change LoyDahua Lin

Cascade is a classic yet powerful architecture that has boosted performance on various tasks. However, how to introduce cascade to instance segmentation remains an open question... (read more)

PDF Abstract

Evaluation results from the paper


 SOTA for Instance Segmentation on COCO (using extra training data)

     Get a GitHub badge
Task Dataset Model Metric name Metric value Global rank Uses extra
training data
Compare
Instance Segmentation COCO HTC + ResNeXt-101-FPN Average Precision 41.2% # 3
Instance Segmentation COCO HTC + ResNeXt-101-FPN + DCN Average Precision 43.9% # 1
Object Detection COCO minival HTC (cascade) box AP 43.2 # 14
Object Detection COCO minival HTC (cascade) AP50 59.4 # 21
Object Detection COCO minival HTC (cascade) AP75 40.7 # 28
Object Detection COCO minival HTC (cascade) APS 20.3 # 25
Object Detection COCO minival HTC (cascade) APM 40.9 # 26
Object Detection COCO minival HTC (cascade) APL 52.3 # 18
Object Detection COCO test-dev HTC (ResNeXt-101-FPN) box AP 47.1 # 5
Object Detection COCO test-dev HTC (ResNeXt-101-FPN) AP50 63.9 # 20
Object Detection COCO test-dev HTC (ResNeXt-101-FPN) AP75 44.7 # 30
Object Detection COCO test-dev HTC (ResNeXt-101-FPN) APS 22.8 # 33
Object Detection COCO test-dev HTC (ResNeXt-101-FPN) APM 43.9 # 29
Object Detection COCO test-dev HTC (ResNeXt-101-FPN) APL 54.6 # 23