{"id":"W4367672600","doi":"10.36227/techrxiv.22133894.v2","title":"Image Captioning for the Visually Impaired and Blind: A Recipe for Low-Resource Languages","year":2023,"lang":"en","type":"preprint","venue":"","topic":"Multimodal Machine Learning Applications","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"ca_institutions":"Artificial Intelligence in Medicine (Canada)","funders":"","keywords":"Closed captioning; Computer science; Visually impaired; Assistive technology; Hearing impaired; Human–computer interaction; Artificial intelligence; Speech recognition; Computer vision; Image (mathematics); Natural language processing; Audiology","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":[0.0009738504,0.001321166,0.000452376,0.0008440503,0.0008847749,0.001709364,0.0008820334,0.001389499,0.01179079],"category_scores_gemma":[0.004399681,0.0002465159,0.0005087951,0.0004253857,0.0009834793,0.002674636,0.001475916,0.001095,0.00584759],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003677573,"about_ca_system_score_gemma":0.0004693846,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0008329105,"about_ca_topic_score_gemma":0.001270349,"domain_scores_codex":[0.9996098,0.0001582876,0.00002874234,0.00005643427,0.0001083187,0.00003848927],"domain_scores_gemma":[0.9982126,0.0005554768,0.00009494138,0.0003613612,0.000633386,0.0001422012],"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":[0.0007487987,0.0002199367,0.001102961,0.002157765,0.00004999853,0.001462033,0.003555383,0.004333822,0.1800559,0.01361347,0.08601889,0.7066811],"study_design_scores_gemma":[0.0001785112,0.001342385,0.008426102,0.001624291,0.0002732173,0.01040261,0.006704146,0.08587207,0.3109618,0.05744722,0.516218,0.0005497058],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.04156177,0.003073598,0.897669,0.005733676,0.001444584,0.001103178,0.001175653,0.01594823,0.03229037],"genre_scores_gemma":[0.2280831,0.00334132,0.7414619,0.001827999,0.0005109404,0.0007358573,0.001640051,0.00169175,0.02070699],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01179079,"threshold_uncertainty_score":0.03944409,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03894768839832212,"score_gpt":0.3599525522228245,"score_spread":0.3210048638245024,"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."}}