{"id":"W4413145364","doi":"10.1109/cvpr52734.2025.00360","title":"VISTA: Enhancing Long-Duration and High-Resolution Video Understanding by VIdeo SpatioTemporal Augmentation","year":2025,"lang":"en","type":"article","venue":"","topic":"Advanced Image Processing Techniques","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Waterloo","funders":"","keywords":"Duration (music); Computer science","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":[0.001051541,0.002268153,0.0007773549,0.001087659,0.0003342732,0.001158828,0.001760688,0.00148302,0.004047789],"category_scores_gemma":[0.005516327,0.0003298692,0.001175941,0.0007235287,0.0004271678,0.002837956,0.001558186,0.002156557,0.00213279],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000728864,"about_ca_system_score_gemma":0.0008597636,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.007550417,"about_ca_topic_score_gemma":0.01221344,"domain_scores_codex":[0.9993482,0.0001613996,0.00002844396,0.0002879403,0.0001182999,0.00005565647],"domain_scores_gemma":[0.9987802,0.0005213549,0.0001051523,0.0002766639,0.0002346351,0.00008193659],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.00118044,0.0007652607,0.003533035,0.001037984,0.0003941236,0.0003374663,0.0004322917,0.1081127,0.09720846,0.003059722,0.04300063,0.7409378],"study_design_scores_gemma":[0.0000740028,0.0006515145,0.002835511,0.0001020617,0.0001185164,0.000329043,0.0002623725,0.9248871,0.045386,0.004771194,0.02051502,0.00006756635],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1500174,0.006088185,0.7744102,0.001244694,0.000746248,0.0009499054,0.01196946,0.04495452,0.009619352],"genre_scores_gemma":[0.3974005,0.001709199,0.5575121,0.001056527,0.0002948074,0.0009140458,0.02986155,0.001435947,0.009815249],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.007550417,"threshold_uncertainty_score":0.01501292,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01602883848453157,"score_gpt":0.2835980329974205,"score_spread":0.2675691945128889,"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."}}