{"corpus_id":248987022,"paper_sha":"5f76de56a59744eb1b771e5940cec65dc488af7e","doi":"10.48550/arXiv.2205.10766","arxiv_id":"2205.10766","pmid":null,"pmcid":null,"mag_id":null,"dblp_id":"journals/corr/abs-2205-10766","acl_id":null,"title":"Recent Advances in Embedding Methods for Multi-Object Tracking: A Survey","year":2022,"publication_date":"2022-05-22","venue":"arXiv.org","journal":{"name":"ArXiv","pages":null,"volume":"abs/2205.10766"},"journal_issn":null,"journal_title":null,"publication_types":["JournalArticle","Review"],"pubmed_pub_types":null,"s2_fields_of_study":["Computer Science"],"reference_count":338,"citation_count":24,"influential_citation_count":2,"is_open_access":true,"arxiv_categories":["cs.CV"],"arxiv_license":"http://creativecommons.org/licenses/by/4.0/","arxiv_journal_ref":null,"mesh_headings":null,"chemicals":null,"comments_corrections":null,"source_flags":1,"s2_open_access_pdf_url":"https://arxiv.org/pdf/2205.10766","s2_open_access_landing_url":"https://www.semanticscholar.org/paper/5f76de56a59744eb1b771e5940cec65dc488af7e","s2_open_access_license":null,"s2_open_access_status":"GREEN","pmc_open_access_pdf_url":null,"pmc_open_access_landing_url":null,"pmc_open_access_license":null,"pmc_open_access_status":null,"unpaywall_open_access_pdf_url":null,"unpaywall_open_access_landing_url":null,"unpaywall_open_access_license":null,"unpaywall_open_access_status":null,"abstract":"Multi-object tracking (MOT) aims to associate target objects across video frames in order to obtain entire moving trajectories. With the advancement of deep neural networks and the increasing demand for intelligent video analysis, MOT has gained significantly increased interest in the computer vision community. Embedding methods play an essential role in object location estimation and temporal identity association in MOT. Unlike other computer vision tasks, such as image classification, object detection, re-identification, and segmentation, embedding methods in MOT have large variations, and they have never been systematically analyzed and summarized. In this survey, we first conduct a comprehensive overview with in-depth analysis for embedding methods in MOT from seven different perspectives, including patch-level embedding, single-frame embedding, cross-frame joint embedding, correlation embedding, sequential embedding, tracklet embedding, and cross-track relational embedding. We further summarize the existing widely used MOT datasets and analyze the advantages of existing state-of-the-art methods according to their embedding strategies. Finally, some critical yet under-investigated areas and future research directions are discussed.","claims":[{"public_id":"cl_7780ab9440bc2301025994c34cfeb638","status":"active","text":"Critical under-investigated areas and future research directions in embedding methods for multi-object tracking are identified and discussed.","confidence":0.9,"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/claims/cl_7780ab9440bc2301025994c34cfeb638"},{"public_id":"cl_f30b262f544e2640b4b4fe4aa971d068","status":"active","text":"Embedding methods in multi-object tracking are systematically analyzed and summarized from seven perspectives: patch-level, single-frame, cross-frame joint, correlation, sequential, tracklet, and cross-track relational embedding.","confidence":0.98,"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/claims/cl_f30b262f544e2640b4b4fe4aa971d068"},{"public_id":"cl_b5973d7bcb050ec2c7d2a2218cd67a36","status":"active","text":"Existing multi-object tracking datasets are summarized together with an analysis of the advantages of state-of-the-art methods according to their embedding strategies.","confidence":0.92,"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/claims/cl_b5973d7bcb050ec2c7d2a2218cd67a36"}],"concepts":[{"public_id":"co_12cff1e2f3f5fe341f0ebb43b320296f","status":"active","name":"single-frame embedding","description":"An embedding strategy computed from information within one video frame for tracking.","types":["embedding strategy"],"aliases":[],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_12cff1e2f3f5fe341f0ebb43b320296f"},{"public_id":"co_16282103068e8c19bd4dac1b8359644f","status":"active","name":"correlation embedding","description":"An embedding strategy that models similarity or correspondence relationships for tracking.","types":["embedding strategy"],"aliases":[],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_16282103068e8c19bd4dac1b8359644f"},{"public_id":"co_311a4e231c7e5084d443b5cc034e21b7","status":"active","name":"seven perspectives","description":"The seven analytic categories used to organize the survey of embedding methods.","types":["taxonomy"],"aliases":[],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_311a4e231c7e5084d443b5cc034e21b7"},{"public_id":"co_4a3189cb4dca48ce4040dcb982d1ead1","status":"active","name":"multi-object tracking datasets","description":"Benchmark datasets used to train or evaluate multi-object tracking methods.","types":["dataset","benchmark"],"aliases":["MOT datasets"],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_4a3189cb4dca48ce4040dcb982d1ead1"},{"public_id":"co_54db415ac5a46694f7f2a49e18f0e1be","status":"active","name":"tracklet embedding","description":"An embedding strategy based on short trajectory fragments used in multi-object tracking.","types":["embedding strategy"],"aliases":[],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_54db415ac5a46694f7f2a49e18f0e1be"},{"public_id":"co_6e592aeef2e8344d462bb7d7b8b0e7bf","status":"active","name":"cross-frame joint embedding","description":"An embedding strategy that jointly represents information from multiple frames.","types":["embedding strategy"],"aliases":[],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_6e592aeef2e8344d462bb7d7b8b0e7bf"},{"public_id":"co_7263aa7fdca0fa0cae8b94e6ea2f8fc7","status":"active","name":"future research directions","description":"Open areas and under-explored topics identified for further work in multi-object tracking embeddings.","types":["research direction"],"aliases":[],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_7263aa7fdca0fa0cae8b94e6ea2f8fc7"},{"public_id":"co_875bc322fe8d4a8063f10c4a51598728","status":"active","name":"multi-object tracking","description":"A computer vision task that associates target objects across video frames to form trajectories over time.","types":["task"],"aliases":["MOT"],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_875bc322fe8d4a8063f10c4a51598728"},{"public_id":"co_a9dd48a8d2312cbdb6a40633fffca529","status":"active","name":"sequential embedding","description":"An embedding strategy that uses temporal ordering or sequence information for tracking.","types":["embedding strategy"],"aliases":[],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_a9dd48a8d2312cbdb6a40633fffca529"},{"public_id":"co_d42077c81123dd1287c5561944823607","status":"active","name":"patch-level embedding","description":"An embedding strategy that represents local image patches for tracking-related processing.","types":["embedding strategy"],"aliases":[],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_d42077c81123dd1287c5561944823607"},{"public_id":"co_dea27f9fa38c7709c6944ef11482949f","status":"active","name":"embedding strategies","description":"Different ways embedding representations are designed and applied within multi-object tracking methods.","types":["methodological approach"],"aliases":[],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_dea27f9fa38c7709c6944ef11482949f"},{"public_id":"co_e2f79ac3005083935328aad0231912bd","status":"active","name":"cross-track relational embedding","description":"An embedding strategy that represents relationships between different tracks or tracklets.","types":["embedding strategy"],"aliases":[],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_e2f79ac3005083935328aad0231912bd"},{"public_id":"co_f4b0df27794fcdcc274d5568a90847ec","status":"active","name":"state-of-the-art methods","description":"The leading published methods currently used for comparison in the survey.","types":["methods"],"aliases":["SOTA methods"],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_f4b0df27794fcdcc274d5568a90847ec"},{"public_id":"co_ff5054e54fbc1e83e9a124a32a65af2d","status":"active","name":"embedding methods","description":"Representation-learning approaches used in multi-object tracking to support object localization and temporal identity association.","types":["method"],"aliases":[],"contributors":[{"id":1,"public_id":"12632b8b5f","public_label":"Anonymous (12632b8b5f)","roles":["extraction"],"url":"https://sah.borca.ai/u/12632b8b5f"}],"url":"https://sah.borca.ai/concepts/co_ff5054e54fbc1e83e9a124a32a65af2d"}],"external_ids":{"DOI":"10.48550/arXiv.2205.10766","ArXiv":"2205.10766","PubMed":null,"PubMedCentral":null,"MAG":null,"DBLP":"journals/corr/abs-2205-10766","ACL":null},"open_access":{"is_open_access":true,"pdf_url":"https://arxiv.org/pdf/2205.10766","landing_url":"https://www.semanticscholar.org/paper/5f76de56a59744eb1b771e5940cec65dc488af7e","source":"semantic_scholar","pdf_url_source":"semantic_scholar_open_access_pdf","license":null,"status":"GREEN","reason":null},"reference_availability":{"status":"available","references_indexed":true,"full_text_available":true,"full_text_source":"arxiv","count_basis":"semantic_scholar_metadata","extraction_status":"not_applicable","reason":null},"source":{"provider":"episteme2","base_corpus":"semantic_scholar_dump","freshness_mode":"unknown","basis":["semantic_scholar_metadata","postgres_metadata"],"limits":["paper metadata is based on indexed upstream scholarly datasets","claims and concepts are available only for extracted papers","absence of claims or concepts means no extracted graph data is available in this response"],"status":"available","degraded":false,"degraded_reasons":[],"diagnostics":{"status":"available","degraded":false,"degraded_reasons":[],"metadata_status":"available","graph_status":"available","abstract_status":"available"},"source_flags":1},"paper_id":633385,"paper_uid":"15644eb6-54bf-4f28-b36b-ef1db453e07f","canonical_identity":{"paper_id":633385,"paper_uid":"15644eb6-54bf-4f28-b36b-ef1db453e07f","identity_status":"available","lookup_basis":"semantic_scholar_external_id","compatibility_path":"corpus_id"},"url":"https://sah.borca.ai/papers/248987022"}