{"id":"W3162126217","doi":"10.48550/arxiv.2105.08665","title":"A multimodal deep learning framework for scalable content based visual media retrieval","year":2021,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Advanced Image and Video Retrieval Techniques","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; Scalability; Modular design; Deep learning; Metric (unit); Artificial intelligence; Content-based image retrieval; Image retrieval; Information retrieval; Image (mathematics); Database","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":[0.0006819525,0.0008060544,0.0007937086,0.001031765,0.0003754475,0.001077422,0.001616753,0.001132786,0.004832922],"category_scores_gemma":[0.001710085,0.0003178123,0.0006628484,0.001067872,0.0005006737,0.002170073,0.001770608,0.001523189,0.002284527],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001029638,"about_ca_system_score_gemma":0.0009452412,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.006303659,"about_ca_topic_score_gemma":0.008442577,"domain_scores_codex":[0.9996653,0.00005555257,0.00001730494,0.00008610028,0.0001124359,0.00006343788],"domain_scores_gemma":[0.9997371,0.00005839294,0.00002583347,0.000060996,0.00008955611,0.00002814122],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0002977948,0.0002907354,0.000866367,0.0002489731,0.0001719924,0.0001536684,0.0001095804,0.133801,0.05480093,0.02754448,0.02358005,0.7581344],"study_design_scores_gemma":[0.00001538962,0.00005938841,0.0001913315,0.00001592277,0.00002362722,0.00005590839,0.00001770214,0.9752876,0.007290087,0.01272599,0.004303053,0.00001400438],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.007182632,0.0005409326,0.9869125,0.0002976598,0.00006621349,0.00007221437,0.0003420627,0.002438985,0.002146864],"genre_scores_gemma":[0.3425908,0.0009142681,0.6377465,0.0007325229,0.0002759042,0.0003222914,0.00210839,0.0004532516,0.01485614],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.006303659,"threshold_uncertainty_score":0.01616776,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.08949241081933848,"score_gpt":0.2454015779684627,"score_spread":0.1559091671491243,"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."}}