{"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/predicting-driver-attention-in-critical","title":"Predicting Driver Attention in Critical Situations","arxiv_id":"1711.06406","date":"2017-11-17","proceeding":null,"authors":["Ye Xia","Danqing Zhang","Jinkyu Kim","Ken Nakayama","Karl Zipser","David Whitney"],"abstract":"Robust driver attention prediction for critical situations is a challenging\ncomputer vision problem, yet essential for autonomous driving. Because critical\ndriving moments are so rare, collecting enough data for these situations is\ndifficult with the conventional in-car data collection protocol---tracking eye\nmovements during driving. Here, we first propose a new in-lab driver attention\ncollection protocol and introduce a new driver attention dataset, Berkeley\nDeepDrive Attention (BDD-A) dataset, which is built upon braking event videos\nselected from a large-scale, crowd-sourced driving video dataset. We further\npropose Human Weighted Sampling (HWS) method, which uses human gaze behavior to\nidentify crucial frames of a driving dataset and weights them heavily during\nmodel training. With our dataset and HWS, we built a driver attention\nprediction model that outperforms the state-of-the-art and demonstrates\nsophisticated behaviors, like attending to crossing pedestrians but not giving\nfalse alarms to pedestrians safely walking on the sidewalk. Its prediction\nresults are nearly indistinguishable from ground-truth to humans. Although only\nbeing trained with our in-lab attention data, the model also predicts in-car\ndriver attention data of routine driving with state-of-the-art accuracy. This\nresult not only demonstrates the performance of our model but also proves the\nvalidity and usefulness of our dataset and data collection protocol.","url_abs":"http://arxiv.org/abs/1711.06406v3","url_pdf":"http://arxiv.org/pdf/1711.06406v3.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":"predicting-driver-attention-in-critical","repo_url":"https://github.com/pascalxia/driver_attention_prediction","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"unanswered"}},{"paper_slug":"predicting-driver-attention-in-critical","repo_url":"https://github.com/spirka3/master-cnn","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"gone","observed_at":"2026-09-17","how":"tree_404+repo_404"}}],"tasks":[{"task_slug":"autonomous-driving","task_name":"Autonomous Driving"},{"task_slug":"driver-attention-monitoring","task_name":"Driver Attention Monitoring"}],"methods":[],"datasets_introduced":[{"slug":"bdd-a","name":"BDD-A","full_name":"Berkeley DeepDrive Attention"}],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/1711.06406","atlas_url":"https://app.syntology.ai/?focus=1711.06406","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}