{"id":28669745,"url":"https://github.com/adeboyeml/imageclef_concept_detection","last_synced_at":"2025-07-17T01:36:01.502Z","repository":{"id":294505025,"uuid":"283928836","full_name":"AdeboyeML/Imageclef_Concept_Detection","owner":"AdeboyeML","description":"The aim of this task is to automatically select medical concepts related to each image, as a first step towards generating image captions, medical reports, or to help in medical diagnosis.","archived":false,"fork":false,"pushed_at":"2020-09-06T21:35:47.000Z","size":16711,"stargazers_count":2,"open_issues_count":1,"forks_count":1,"subscribers_count":0,"default_branch":"master","last_synced_at":"2025-06-13T17:43:50.594Z","etag":null,"topics":["cnn-multilabel-classification","imageclef","imageclef-2019","imageclef-concept-detection"],"latest_commit_sha":null,"homepage":"","language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":null,"status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/AdeboyeML.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":null,"code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null}},"created_at":"2020-07-31T03:03:35.000Z","updated_at":"2023-10-31T09:39:11.000Z","dependencies_parsed_at":"2025-06-13T17:34:27.795Z","dependency_job_id":null,"html_url":"https://github.com/AdeboyeML/Imageclef_Concept_Detection","commit_stats":null,"previous_names":["adeboyeml/imageclef_concept_detection"],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/AdeboyeML/Imageclef_Concept_Detection","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/AdeboyeML%2FImageclef_Concept_Detection","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/AdeboyeML%2FImageclef_Concept_Detection/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/AdeboyeML%2FImageclef_Concept_Detection/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/AdeboyeML%2FImageclef_Concept_Detection/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/AdeboyeML","download_url":"https://codeload.github.com/AdeboyeML/Imageclef_Concept_Detection/tar.gz/refs/heads/master","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/AdeboyeML%2FImageclef_Concept_Detection/sbom","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":265558745,"owners_count":23787973,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["cnn-multilabel-classification","imageclef","imageclef-2019","imageclef-concept-detection"],"created_at":"2025-06-13T17:30:23.334Z","updated_at":"2025-07-17T01:36:01.489Z","avatar_url":"https://github.com/AdeboyeML.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"\n\n# Imageclef_Concept_Detection\nThe aim of this task is to automatically detect medical concepts related to each image, as a first step towards generating image captions, medical reports, or to help in medical diagnosis.\n\n### Steps Taken:\n\n- Acquisition of Datasets and Extraction of Images from Tarfiles\n- Data Exploration\n- Data Analysis\n- Data Visualization \n- Data Preprocessing\n- Implementation of Machine learning models \n- Evaluation and Prediction --\n\n\n\n\n### -- Summary (models still needs further training...more compute power required)\n\n### -- Full ROCO (Radiology Objects in COntext) Dataset\nNo | Datasets | No of images\n--- | --- | ---\n0 | Train Dataset | 60963\n1 | Validation Dataset | 7,703\n2 | Test Dataset | 7,662\n3 | Total | 76328\n\n\n- Evaluation metric == F1 Score: is the most suited for imbalanced class labels (in our case -- concepts to be detected).\n\n\n##### - Decision Threshold was tuned on validation dataset, the best threshold was 0.1\n\nNo | Model Description | Dev. f1 Score | Test f1 Score\n--- | --- | --- | ---\n0 | DenseNet-121 Encoder + FFNN (AUEB NLP Group, 2019) | 0.157 | 0.146\n1 | DenseNet-121 Encoder + k-NN Image Retrieval (AUEB NLP Group, 2019) | 0.147 | 0.142\n\n\n\n\n### -- Reduced Dataset\nNo | Datasets | No of images\n--- | --- | ---\n0 | Train Dataset | 30000\n1 | Validation Dataset | 3500\n2 | Test Dataset | 3500\n3 | Total | 37000\n\n\n### -- Summary ( All models still needs retraining)\n\n##### - Decision Threshold was tuned on validation dataset, the best threshold was 0.1\n\nNo | Model Description | Dev. f1 Score | Test f1 Score\n--- | --- | --- | ---\n0 | DenseNet-121 Encoder + FFNN (AUEB NLP Group, 2019) | 0.168 | 0.161\n1 | Resnet 101 + FFNN, Multi-label classification in Xu, et al 2019 | 0.168 | 0.160\n2 | DenseNet-121 Encoder + k-NN Image Retrieval (AUEB NLP Group, 2019) | 0.150 | 0.142\n4 | ResNet 101 + Data Filtering (Df1) -- Xu et al., 2019 (Damo Group) | 0.169 | 0.160\n5 | ResNet 101 + Data Filtering (Df3) -- Xu et al., 2019 (Damo Group) | 0.170 | 0.163\n\n\n\n### Python Scripts\n- Download ROCO tar files and extract images from the files -\u003e `download_extract.py,`\n- DenseNet-121 Encoder/Resnet 101 + Feed Forward Neural Network -\u003e `train_model_get_threshold.py,`\n- DenseNet-121 Encoder + k-NN Image Retrieval -\u003e `knn_train_test.py,`\n- ResNet 101 + Data Filtering (Df1/Df3) -\u003e `filtered_model.py,`\n- make predictions on test data -\u003e `make_predictions.py,`\n\n\n\n### Scientific papers References\n\n[AUEB NLP Group, 2019](http://ceur-ws.org/Vol-2380/paper_136.pdf)\n\n[Damo Group, 2019](http://ceur-ws.org/Vol-2380/paper_141.pdf)\n\n[Pelka et al., 2019](http://ceur-ws.org/Vol-2380/paper_245.pdf)\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fadeboyeml%2Fimageclef_concept_detection","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fadeboyeml%2Fimageclef_concept_detection","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fadeboyeml%2Fimageclef_concept_detection/lists"}