{"url":"/method/cascade-mask-r-cnn","slug":"cascade-mask-r-cnn","name":"Cascade Mask R-CNN","full_name":"Cascade Mask R-CNN","full_name_withheld":false,"description_markdown":"**Cascade Mask R-CNN** extends [Cascade R-CNN](https://paperswithcode.com/method/cascade-r-cnn) to instance segmentation, by adding a\r\nmask head to the cascade.\r\n\r\nIn the [Mask R-CNN](https://paperswithcode.com/method/mask-r-cnn), the segmentation branch is inserted in parallel to the detection branch. However, the Cascade [R-CNN](https://paperswithcode.com/method/r-cnn) has multiple detection branches. This raises the questions of 1) where to add the segmentation branch and 2) how many segmentation branches to add. The authors consider three strategies for mask prediction in the Cascade R-CNN. The first two strategies address the first question, adding a single mask prediction head at either the first or last stage of the Cascade R-CNN. Since the instances used to train the segmentation branch are the positives of the detection branch, their number varies in these two strategies. Placing the segmentation head later on the cascade leads to more examples. However, because segmentation is a pixel-wise operation, a large number of highly overlapping instances is not necessarily as helpful as for object detection, which is a patch-based operation. The third strategy addresses the second question, adding a segmentation branch to each\r\ncascade stage. This maximizes the diversity of samples used to learn the mask prediction task. \r\n\r\nAt inference time, all three strategies predict the segmentation masks on the patches produced by the final object detection stage, irrespective of the cascade stage on which the segmentation mask is implemented and how many segmentation branches there are.","description_state":"present","introduced_year":null,"introduced_by":{"title":"Cascade R-CNN: Delving into High Quality Object Detection","paper":"/paper/cascade-r-cnn-delving-into-high-quality","first_author":"Zhaowei Cai","n_authors":2,"url_abs":null,"archive_paper_url":"https://paperswithcode.com/paper/cascade-r-cnn-delving-into-high-quality"},"source":{"url":"http://arxiv.org/abs/1712.00726v1","title":"Cascade R-CNN: Delving into High Quality Object Detection","url_on_a_paper_host":true},"code_snippet_url":null,"code_snippet_url_on_a_code_host":false,"categories":[{"area":"Computer Vision","area_id":"computer-vision","collection":"Instance Segmentation Models","url":"/methods/category/instance-segmentation-models","pwc_aliases":[]}],"n_papers_tagged":23,"archive_num_papers":23,"papers_newest_first":[{"paper":null,"title":"RipVIS: Rip Currents Video Instance Segmentation Benchmark for Beach Monitoring and Safety","date":"2025-04-01","arxiv_id":"2504.01128","n_code_links":0,"syntology":null},{"paper":"/paper/overlock-an-overview-first-look-closely-next","title":"OverLoCK: An Overview-first-Look-Closely-next ConvNet with Context-Mixing Dynamic Kernels","date":"2025-02-27","arxiv_id":"2502.20087","n_code_links":1,"syntology":{"ran":0,"of":3,"unverified":3,"pointer_only":0}},{"paper":null,"title":"Hierarchical Side-Tuning for Vision Transformers","date":"2023-10-09","arxiv_id":"2310.05393","n_code_links":0,"syntology":null},{"paper":null,"title":"DMKD: Improving Feature-based Knowledge Distillation for Object Detection Via Dual Masking Augmentation","date":"2023-09-06","arxiv_id":"2309.02719","n_code_links":0,"syntology":null},{"paper":null,"title":"Non-Hierarchical Transformers for Pedestrian Segmentation","date":"2023-07-11","arxiv_id":"2311.02506","n_code_links":0,"syntology":null},{"paper":null,"title":"BiViT: Extremely Compressed Binary Vision Transformers","date":"2023-01-01","arxiv_id":null,"n_code_links":0,"syntology":null},{"paper":null,"title":"BiViT: Extremely Compressed Binary Vision Transformer","date":"2022-11-14","arxiv_id":"2211.07091","n_code_links":0,"syntology":null},{"paper":"/paper/a-tri-layer-plugin-to-improve-occluded","title":"A Tri-Layer Plugin to Improve Occluded Detection","date":"2022-10-18","arxiv_id":"2210.10046","n_code_links":1,"syntology":null},{"paper":"/paper/fq-vit-fully-quantized-vision-transformer","title":"FQ-ViT: Post-Training Quantization for Fully Quantized Vision Transformer","date":"2021-11-27","arxiv_id":"2111.13824","n_code_links":1,"syntology":{"ran":3,"of":3,"unverified":0,"pointer_only":1}},{"paper":null,"title":"VTLayout: Fusion of Visual and Text Features for Document Layout Analysis","date":"2021-08-12","arxiv_id":"2108.13297","n_code_links":0,"syntology":null},{"paper":"/paper/k-net-towards-unified-image-segmentation","title":"K-Net: Towards Unified Image Segmentation","date":"2021-06-28","arxiv_id":"2106.14855","n_code_links":1,"syntology":{"ran":0,"of":1,"unverified":1,"pointer_only":0}},{"paper":null,"title":"A2-FPN: Attention Aggregation Based Feature Pyramid Network for Instance Segmentation","date":"2021-06-19","arxiv_id":null,"n_code_links":0,"syntology":null},{"paper":"/paper/tncr-table-net-detection-and-classification","title":"TNCR: Table Net Detection and Classification Dataset","date":"2021-06-19","arxiv_id":"2106.15322","n_code_links":1,"syntology":null},{"paper":null,"title":"A^2-FPN: Attention Aggregation based Feature Pyramid Network for Instance Segmentation","date":"2021-05-07","arxiv_id":"2105.03186","n_code_links":0,"syntology":null},{"paper":"/paper/queryinst-parallelly-supervised-mask-query","title":"Instances as Queries","date":"2021-05-05","arxiv_id":"2105.01928","n_code_links":5,"syntology":{"ran":1,"of":3,"unverified":2,"pointer_only":0}},{"paper":null,"title":"Object Detection for Understanding Assembly Instruction Using Context-aware Data Augmentation and Cascade Mask R-CNN","date":"2021-01-07","arxiv_id":"2101.02509","n_code_links":0,"syntology":null},{"paper":"/paper/scnet-training-inference-sample-consistency","title":"SCNet: Training Inference Sample Consistency for Instance Segmentation","date":"2020-12-18","arxiv_id":"2012.10150","n_code_links":2,"syntology":null},{"paper":"/paper/simple-copy-paste-is-a-strong-data","title":"Simple Copy-Paste is a Strong Data Augmentation Method for Instance Segmentation","date":"2020-12-13","arxiv_id":"2012.07177","n_code_links":5,"syntology":{"ran":2,"of":2,"unverified":0,"pointer_only":0}},{"paper":"/paper/cascadetabnet-an-approach-for-end-to-end","title":"CascadeTabNet: An approach for end to end table detection and structure recognition from image-based documents","date":"2020-04-27","arxiv_id":"2004.12629","n_code_links":3,"syntology":{"ran":0,"of":2,"unverified":2,"pointer_only":0}},{"paper":"/paper/cbnet-a-novel-composite-backbone-network","title":"CBNet: A Novel Composite Backbone Network Architecture for Object Detection","date":"2019-09-09","arxiv_id":"1909.03625","n_code_links":6,"syntology":{"ran":0,"of":4,"unverified":4,"pointer_only":1}},{"paper":"/paper/gcnet-non-local-networks-meet-squeeze","title":"GCNet: Non-local Networks Meet Squeeze-Excitation Networks and Beyond","date":"2019-04-25","arxiv_id":"1904.11492","n_code_links":9,"syntology":null},{"paper":"/paper/hybrid-task-cascade-for-instance-segmentation","title":"Hybrid Task Cascade for Instance Segmentation","date":"2019-01-22","arxiv_id":"1901.07518","n_code_links":5,"syntology":{"ran":0,"of":11,"unverified":11,"pointer_only":0}},{"paper":"/paper/cascade-r-cnn-delving-into-high-quality","title":"Cascade R-CNN: Delving into High Quality Object Detection","date":"2017-12-03","arxiv_id":"1712.00726","n_code_links":8,"syntology":{"ran":2,"of":2,"unverified":0,"pointer_only":2}}],"papers_shown":23,"tasks":[{"task":"/task/instance-segmentation","name":"Instance Segmentation","papers":14},{"task":"/task/object-detection","name":"Object Detection","papers":14},{"task":"/task/semantic-segmentation","name":"Semantic Segmentation","papers":12},{"task":"/task/object-detection-1","name":"object-detection","papers":10},{"task":"/task/segmentation","name":"Segmentation","papers":8},{"task":"/task/object","name":"Object","papers":3},{"task":"/task/binarization","name":"Binarization","papers":2},{"task":"/task/data-augmentation","name":"Data Augmentation","papers":2},{"task":"/task/image-augmentation","name":"Image Augmentation","papers":2},{"task":"/task/image-classification","name":"Image Classification","papers":2},{"task":"/task/table-detection","name":"Table Detection","papers":2},{"task":"/task/transfer-learning","name":"Transfer Learning","papers":2},{"task":"/task/video-instance-segmentation","name":"Video Instance Segmentation","papers":2},{"task":"/task/2d-object-detection","name":"2D Object Detection","papers":1},{"task":"/task/classification-1","name":"Classification","papers":1},{"task":"/task/deep-learning","name":"Deep Learning","papers":1},{"task":"/task/diversity","name":"Diversity","papers":1},{"task":"/task/document-layout-analysis","name":"Document Layout Analysis","papers":1},{"task":"/task/image-segmentation","name":"Image Segmentation","papers":1},{"task":"/task/knowledge-distillation","name":"Knowledge Distillation","papers":1}],"tasks_shown":20,"n_tasks":28,"usage_by_year":[{"year":"2017","papers":1},{"year":"2019","papers":3},{"year":"2020","papers":3},{"year":"2021","papers":8},{"year":"2022","papers":2},{"year":"2023","papers":4},{"year":"2025","papers":2}],"row_source":"methods_table","archive":{"source":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","archive_url":"https://paperswithcode.com/method/cascade-mask-r-cnn"},"syntology_read_at":"2026-09-24T18:15:14+00:00"}