{"id":"W4402904107","doi":"10.1109/cvprw63382.2024.00708","title":"TrafficVLM: A Controllable Visual Language Model for Traffic Video Captioning","year":2024,"lang":"en","type":"article","venue":"","topic":"Multimodal Machine Learning Applications","field":"Computer Science","cited_by":21,"is_retracted":false,"has_abstract":true,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Closed captioning; Computer science; Speech recognition; Artificial intelligence; Image (mathematics)","routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005875469,0.001067513,0.0005591543,0.0008025985,0.0003054919,0.00096113,0.002381353,0.001003388,0.004504371],"category_scores_gemma":[0.002519187,0.0005321585,0.001212187,0.000557137,0.0005254351,0.001546794,0.001536898,0.001534829,0.001881844],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001020112,"about_ca_system_score_gemma":0.0008672308,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.008989243,"about_ca_topic_score_gemma":0.01391124,"domain_scores_codex":[0.9996397,0.00007588274,0.00001635935,0.0001330255,0.00009017324,0.00004474346],"domain_scores_gemma":[0.9995353,0.0001784719,0.00004091855,0.00007784647,0.0001241502,0.00004336665],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0004637541,0.0002828421,0.001792804,0.0003512103,0.0001251689,0.0003822888,0.0002965261,0.5302443,0.0291135,0.02309627,0.03711463,0.3767367],"study_design_scores_gemma":[0.000009709384,0.00001857337,0.00009853641,0.000006547014,0.000006340242,0.00003406704,0.000009047614,0.9917761,0.001921843,0.003597254,0.002513073,0.000008955655],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.008736009,0.0002406622,0.9771871,0.0001744684,0.00008436661,0.0001511176,0.001303402,0.01014538,0.001977581],"genre_scores_gemma":[0.4020324,0.0005704785,0.5759635,0.0006375075,0.000125473,0.0008915401,0.007392135,0.002261359,0.01012558],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.008989243,"threshold_uncertainty_score":0.01787382,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01249358995875663,"score_gpt":0.3133651745960067,"score_spread":0.3008715846372501,"validation_status":"score_only:v0-immature-baseline","note":"Baseline scores from an immature model (maturity gate not passed). Scores rank; they never assert a category."}}