Parallel Detection-and-Segmentation Learning for Weakly Supervised Instance Segmentation

Parallel Detection-and-Segmentation Learning for Weakly Supervised Instance Segmentation. Shen, Y., Cao, L., Chen, Z., Zhang, B., Su, C., Wu, Y., Huang, F., & Ji, R. 2021.

Paper abstract bibtex

Weakly supervised instance segmentation (WSIS) with only image-level labels has recently drawn much attention. To date, bottom-up WSIS methods refine discrimi-native cues from classifiers with sophisticated multi-stage training procedures, which also suffer from inconsistent object boundaries. And top-down WSIS methods are formulated as cascade detection-to-segmentation pipeline, in which the quality of segmentation learning heavily depends on pseudo masks generated from detectors. In this paper, we propose a unified parallel detection-and-segmentation learning (PDSL) framework to learn instance segmentation with only image-level labels, which draws inspiration from both top-down and bottom-up instance segmentation approaches. The detection module is the same as the typical design of any weakly supervised object detection, while the segmentation module leverages self-supervised learning to model class-agnostic foreground extraction, following by self-training to refine class-specific segmentation. We further design instance-activation correlation module to improve the coherence between detection and segmentation branches. Extensive experiments verify that the proposed method outperforms baselines and achieves the state-of-the-art results on PASCAL VOC and MS COCO.

@misc{
 title = {Parallel Detection-and-Segmentation Learning for Weakly Supervised Instance Segmentation},
 type = {misc},
 year = {2021},
 pages = {8198-8208},
 id = {b212b411-abd5-3f3e-bc0f-e751d652326a},
 created = {2023-11-07T08:32:19.145Z},
 accessed = {2023-11-07},
 file_attached = {true},
 profile_id = {f1f70cad-e32d-3de2-a3c0-be1736cb88be},
 group_id = {5ec9cc91-a5d6-3de5-82f3-3ef3d98a89c1},
 last_modified = {2023-11-07T08:32:24.535Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {false},
 hidden = {false},
 folder_uuids = {bc66e353-ef41-46d4-8108-778d5481c126,bd3c6f2e-3514-47cf-bc42-12db8b9abe45},
 private_publication = {false},
 abstract = {Weakly supervised instance segmentation (WSIS) with only image-level labels has recently drawn much attention. To date, bottom-up WSIS methods refine discrimi-native cues from classifiers with sophisticated multi-stage training procedures, which also suffer from inconsistent object boundaries. And top-down WSIS methods are formulated as cascade detection-to-segmentation pipeline, in which the quality of segmentation learning heavily depends on pseudo masks generated from detectors. In this paper, we propose a unified parallel detection-and-segmentation learning (PDSL) framework to learn instance segmentation with only image-level labels, which draws inspiration from both top-down and bottom-up instance segmentation approaches. The detection module is the same as the typical design of any weakly supervised object detection, while the segmentation module leverages self-supervised learning to model class-agnostic foreground extraction, following by self-training to refine class-specific segmentation. We further design instance-activation correlation module to improve the coherence between detection and segmentation branches. Extensive experiments verify that the proposed method outperforms baselines and achieves the state-of-the-art results on PASCAL VOC and MS COCO.},
 bibtype = {misc},
 author = {Shen, Yunhang and Cao, Liujuan and Chen, Zhiwei and Zhang, Baochang and Su, Chi and Wu, Yongjian and Huang, Feiyue and Ji, Rongrong}
}

Downloads: 0

{"_id":"43w4Y6aaAJhmfiT5X","bibbaseid":"shen-cao-chen-zhang-su-wu-huang-ji-paralleldetectionandsegmentationlearningforweaklysupervisedinstancesegmentation-2021","author_short":["Shen, Y.","Cao, L.","Chen, Z.","Zhang, B.","Su, C.","Wu, Y.","Huang, F.","Ji, R."],"bibdata":{"title":"Parallel Detection-and-Segmentation Learning for Weakly Supervised Instance Segmentation","type":"misc","year":"2021","pages":"8198-8208","id":"b212b411-abd5-3f3e-bc0f-e751d652326a","created":"2023-11-07T08:32:19.145Z","accessed":"2023-11-07","file_attached":"true","profile_id":"f1f70cad-e32d-3de2-a3c0-be1736cb88be","group_id":"5ec9cc91-a5d6-3de5-82f3-3ef3d98a89c1","last_modified":"2023-11-07T08:32:24.535Z","read":false,"starred":false,"authored":false,"confirmed":false,"hidden":false,"folder_uuids":"bc66e353-ef41-46d4-8108-778d5481c126,bd3c6f2e-3514-47cf-bc42-12db8b9abe45","private_publication":false,"abstract":"Weakly supervised instance segmentation (WSIS) with only image-level labels has recently drawn much attention. To date, bottom-up WSIS methods refine discrimi-native cues from classifiers with sophisticated multi-stage training procedures, which also suffer from inconsistent object boundaries. And top-down WSIS methods are formulated as cascade detection-to-segmentation pipeline, in which the quality of segmentation learning heavily depends on pseudo masks generated from detectors. In this paper, we propose a unified parallel detection-and-segmentation learning (PDSL) framework to learn instance segmentation with only image-level labels, which draws inspiration from both top-down and bottom-up instance segmentation approaches. The detection module is the same as the typical design of any weakly supervised object detection, while the segmentation module leverages self-supervised learning to model class-agnostic foreground extraction, following by self-training to refine class-specific segmentation. We further design instance-activation correlation module to improve the coherence between detection and segmentation branches. Extensive experiments verify that the proposed method outperforms baselines and achieves the state-of-the-art results on PASCAL VOC and MS COCO.","bibtype":"misc","author":"Shen, Yunhang and Cao, Liujuan and Chen, Zhiwei and Zhang, Baochang and Su, Chi and Wu, Yongjian and Huang, Feiyue and Ji, Rongrong","bibtex":"@misc{\n title = {Parallel Detection-and-Segmentation Learning for Weakly Supervised Instance Segmentation},\n type = {misc},\n year = {2021},\n pages = {8198-8208},\n id = {b212b411-abd5-3f3e-bc0f-e751d652326a},\n created = {2023-11-07T08:32:19.145Z},\n accessed = {2023-11-07},\n file_attached = {true},\n profile_id = {f1f70cad-e32d-3de2-a3c0-be1736cb88be},\n group_id = {5ec9cc91-a5d6-3de5-82f3-3ef3d98a89c1},\n last_modified = {2023-11-07T08:32:24.535Z},\n read = {false},\n starred = {false},\n authored = {false},\n confirmed = {false},\n hidden = {false},\n folder_uuids = {bc66e353-ef41-46d4-8108-778d5481c126,bd3c6f2e-3514-47cf-bc42-12db8b9abe45},\n private_publication = {false},\n abstract = {Weakly supervised instance segmentation (WSIS) with only image-level labels has recently drawn much attention. To date, bottom-up WSIS methods refine discrimi-native cues from classifiers with sophisticated multi-stage training procedures, which also suffer from inconsistent object boundaries. And top-down WSIS methods are formulated as cascade detection-to-segmentation pipeline, in which the quality of segmentation learning heavily depends on pseudo masks generated from detectors. In this paper, we propose a unified parallel detection-and-segmentation learning (PDSL) framework to learn instance segmentation with only image-level labels, which draws inspiration from both top-down and bottom-up instance segmentation approaches. The detection module is the same as the typical design of any weakly supervised object detection, while the segmentation module leverages self-supervised learning to model class-agnostic foreground extraction, following by self-training to refine class-specific segmentation. We further design instance-activation correlation module to improve the coherence between detection and segmentation branches. Extensive experiments verify that the proposed method outperforms baselines and achieves the state-of-the-art results on PASCAL VOC and MS COCO.},\n bibtype = {misc},\n author = {Shen, Yunhang and Cao, Liujuan and Chen, Zhiwei and Zhang, Baochang and Su, Chi and Wu, Yongjian and Huang, Feiyue and Ji, Rongrong}\n}","author_short":["Shen, Y.","Cao, L.","Chen, Z.","Zhang, B.","Su, C.","Wu, Y.","Huang, F.","Ji, R."],"urls":{"Paper":"https://bibbase.org/service/mendeley/bfbbf840-4c42-3914-a463-19024f50b30c/file/7309f4a4-0917-1f6b-6f14-4466a705cb0a/full_text.pdf.pdf"},"biburl":"https://bibbase.org/service/mendeley/bfbbf840-4c42-3914-a463-19024f50b30c","bibbaseid":"shen-cao-chen-zhang-su-wu-huang-ji-paralleldetectionandsegmentationlearningforweaklysupervisedinstancesegmentation-2021","role":"author","metadata":{"authorlinks":{}},"downloads":0},"bibtype":"misc","biburl":"https://bibbase.org/service/mendeley/bfbbf840-4c42-3914-a463-19024f50b30c","dataSources":["2252seNhipfTmjEBQ"],"keywords":[],"search_terms":["parallel","detection","segmentation","learning","weakly","supervised","instance","segmentation","shen","cao","chen","zhang","su","wu","huang","ji"],"title":"Parallel Detection-and-Segmentation Learning for Weakly Supervised Instance Segmentation","year":2021}