{"id":"W7131093438","doi":"10.1109/iccvw69036.2025.00221","title":"Refining Naive Annotations with Limited Expert Guidance for Semantic Segmentation: A Case Study on Underwater Echograms","year":2025,"lang":"","type":"article","venue":"","topic":"Multimodal Machine Learning Applications","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"Fisheries and Oceans Canada; Canadian Water Network; ASL Environmental Sciences (Canada); University of Victoria","funders":"Alliance de recherche numérique du Canada","keywords":"Metadata; Annotation; Ground truth; Context (archaeology); Intersection (aeronautics); Artificial neural network; Segmentation; Convolutional neural network; Domain (mathematical analysis)","routes":{"ca_aff":true,"ca_fund":true,"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":[0.004372505,0.001823088,0.001082918,0.001672862,0.001141634,0.001752974,0.001994027,0.002852452,0.001910917],"category_scores_gemma":[0.01142985,0.0005355648,0.0009439245,0.001284652,0.001607571,0.002360947,0.001943456,0.001820953,0.001501634],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001371817,"about_ca_system_score_gemma":0.001237187,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01441825,"about_ca_topic_score_gemma":0.03560785,"domain_scores_codex":[0.9967957,0.001180598,0.000177357,0.001102076,0.0005266846,0.0002175672],"domain_scores_gemma":[0.9896609,0.007374596,0.0002616076,0.001305247,0.001157332,0.0002403524],"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":[0.001997856,0.0005996453,0.01271186,0.001199347,0.0002491397,0.003221902,0.004648995,0.3122662,0.08618581,0.008052985,0.02580256,0.5430636],"study_design_scores_gemma":[0.00009477259,0.0002351313,0.004522539,0.0001172824,0.0000910922,0.000713377,0.001378982,0.9060962,0.05744755,0.00997643,0.01924766,0.00007896751],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.4086502,0.001098441,0.5673692,0.001096533,0.0001755337,0.0003139747,0.001875139,0.01256052,0.006860353],"genre_scores_gemma":[0.5547087,0.0002916839,0.4331943,0.0003795487,0.00005876175,0.0001397062,0.004888013,0.001637685,0.004701629],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01441825,"threshold_uncertainty_score":0.02866864,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.04203390367071803,"score_gpt":0.3603349623937063,"score_spread":0.3183010587229883,"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."}}