{"id":"W2900626451","doi":"10.1109/cvprw.2018.00260","title":"Image Caption Generation with Hierarchical Contextual Visual Spatial Attention","year":2018,"lang":"en","type":"article","venue":"","topic":"Multimodal Machine Learning Applications","field":"Computer Science","cited_by":25,"is_retracted":false,"has_abstract":true,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Computer science; Leverage (statistics); Artificial intelligence; Grid; Word (group theory); Context (archaeology); Spatial contextual awareness; Layer (electronics); Context model; Image (mathematics); Object (grammar); Pattern recognition (psychology); Computer vision; Natural language processing; Linguistics; Geography","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":[0.0003893855,0.001687162,0.0007331445,0.0006532456,0.0003458415,0.0009139612,0.001997471,0.001446391,0.006835502],"category_scores_gemma":[0.001786879,0.0004572717,0.0009292862,0.0007048089,0.0005553042,0.001950887,0.001435057,0.001749014,0.00232127],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0009347576,"about_ca_system_score_gemma":0.000762417,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.005499506,"about_ca_topic_score_gemma":0.007329881,"domain_scores_codex":[0.9997279,0.00003687401,0.00001260807,0.0001149517,0.00005788695,0.00004964388],"domain_scores_gemma":[0.9996325,0.0001039173,0.00003286509,0.00008877654,0.0001064699,0.000035478],"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":[0.00038674,0.0002555985,0.0006658488,0.0005529808,0.0001697873,0.0005377731,0.0002953344,0.1287986,0.08012357,0.007281221,0.03868695,0.7422456],"study_design_scores_gemma":[0.00003808152,0.0001082789,0.0003817695,0.00003140789,0.00006835462,0.0001795841,0.00004239736,0.9455212,0.03433292,0.009402992,0.009863,0.00003002208],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.03134018,0.002618287,0.9199169,0.0007928479,0.0007671886,0.000272229,0.001184185,0.03132522,0.01178308],"genre_scores_gemma":[0.4977427,0.001386884,0.4773543,0.001322031,0.0003755931,0.0004039792,0.004217874,0.001253246,0.0159434],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.006835502,"threshold_uncertainty_score":0.02286696,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01382938427846726,"score_gpt":0.2899706842396153,"score_spread":0.276141299961148,"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."}}