{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/self-supervised-audio-visual-representation","title":"Self-Supervised Audio-Visual Representation Learning with Relaxed Cross-Modal Synchronicity","arxiv_id":"2111.05329","date":"2021-11-09","proceeding":null,"authors":["Pritam Sarkar","Ali Etemad"],"abstract":"We present CrissCross, a self-supervised framework for learning audio-visual representations. A novel notion is introduced in our framework whereby in addition to learning the intra-modal and standard 'synchronous' cross-modal relations, CrissCross also learns 'asynchronous' cross-modal relationships. We perform in-depth studies showing that by relaxing the temporal synchronicity between the audio and visual modalities, the network learns strong generalized representations useful for a variety of downstream tasks. To pretrain our proposed solution, we use 3 different datasets with varying sizes, Kinetics-Sound, Kinetics400, and AudioSet. The learned representations are evaluated on a number of downstream tasks namely action recognition, sound classification, and action retrieval. Our experiments show that CrissCross either outperforms or achieves performances on par with the current state-of-the-art self-supervised methods on action recognition and action retrieval with UCF101 and HMDB51, as well as sound classification with ESC50 and DCASE. Moreover, CrissCross outperforms fully-supervised pretraining while pretrained on Kinetics-Sound. The codes and pretrained models are available on the project website.","url_abs":"https://arxiv.org/abs/2111.05329v5","url_pdf":"https://arxiv.org/pdf/2111.05329v5.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"self-supervised-audio-visual-representation","repo_url":"https://github.com/pritamqu/CrissCross","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"NOASSERTION"}}],"tasks":[{"task_slug":"audio-classification","task_name":"Audio Classification"},{"task_slug":"retrieval","task_name":"Retrieval"},{"task_slug":"self-supervised-action-recognition","task_name":"Self-Supervised Action Recognition"},{"task_slug":"self-supervised-audio-classification","task_name":"Self-Supervised Audio Classification"},{"task_slug":"self-supervised-learning","task_name":"Self-Supervised Learning"},{"task_slug":"self-supervised-sound-classification","task_name":"Self-Supervised Sound Classification"},{"task_slug":"self-supervised-video-retrieval","task_name":"Self-supervised Video Retrieval"},{"task_slug":"sound-classification","task_name":"Sound Classification"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/audio-classification-on-dcase","task":"Audio Classification","dataset":"DCASE","model":"CrissCross (AudioSet)","rank_in_archive_order":1,"of":5,"metrics":{"PRE-TRAINING DATASET":"AudioSet","Top-1 Accuracy":"97"},"uses_additional_data":false},{"leaderboard":"/sota/audio-classification-on-dcase","task":"Audio Classification","dataset":"DCASE","model":"CrissCross (Kinetics-400)","rank_in_archive_order":2,"of":5,"metrics":{"PRE-TRAINING DATASET":"Kinetics-400","Top-1 Accuracy":"96"},"uses_additional_data":false},{"leaderboard":"/sota/audio-classification-on-dcase","task":"Audio Classification","dataset":"DCASE","model":"CrissCross (Kinetics-Sound)","rank_in_archive_order":5,"of":5,"metrics":{"PRE-TRAINING DATASET":"Kinetics-Sound","Top-1 Accuracy":"93"},"uses_additional_data":false},{"leaderboard":"/sota/self-supervised-action-recognition-on-hmdb51","task":"Self-Supervised Action Recognition","dataset":"HMDB51","model":"CrissCross (AudioSet)","rank_in_archive_order":11,"of":48,"metrics":{"Frozen":"false","Pre-Training Dataset":"AudioSet","Top-1 Accuracy":"66.8"},"uses_additional_data":false},{"leaderboard":"/sota/self-supervised-action-recognition-on-hmdb51","task":"Self-Supervised Action Recognition","dataset":"HMDB51","model":"CrissCross (Kinetics400)","rank_in_archive_order":18,"of":48,"metrics":{"Frozen":"false","Pre-Training Dataset":"Kinetics400","Top-1 Accuracy":"64.7"},"uses_additional_data":false},{"leaderboard":"/sota/self-supervised-action-recognition-on-hmdb51","task":"Self-Supervised Action Recognition","dataset":"HMDB51","model":"CrissCross (Kinetics-Sound)","rank_in_archive_order":26,"of":48,"metrics":{"Frozen":"false","Pre-Training Dataset":"Kinetics-Sound","Top-1 Accuracy":"60.5"},"uses_additional_data":false},{"leaderboard":"/sota/self-supervised-action-recognition-on-ucf101","task":"Self-Supervised Action Recognition","dataset":"UCF101","model":"CrissCross (AudioSet)","rank_in_archive_order":16,"of":53,"metrics":{"3-fold Accuracy":"92.4","Frozen":"false","Pre-Training Dataset":"AudioSet"},"uses_additional_data":false},{"leaderboard":"/sota/self-supervised-action-recognition-on-ucf101","task":"Self-Supervised Action Recognition","dataset":"UCF101","model":"CrissCross (Kinetics400)","rank_in_archive_order":19,"of":53,"metrics":{"3-fold Accuracy":"91.5","Frozen":"false","Pre-Training Dataset":"Kinetics400"},"uses_additional_data":false},{"leaderboard":"/sota/self-supervised-action-recognition-on-ucf101","task":"Self-Supervised Action Recognition","dataset":"UCF101","model":"CrissCross (Kinetics-Sound)","rank_in_archive_order":25,"of":53,"metrics":{"3-fold Accuracy":"88.3","Frozen":"false","Pre-Training Dataset":"Kinetics-Sound"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2111.05329","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}