{"id":19190759,"url":"https://github.com/yashbit/iscr_cnn","last_synced_at":"2025-06-23T08:34:43.947Z","repository":{"id":185832779,"uuid":"658777328","full_name":"YashBit/ISCR_CNN","owner":"YashBit","description":"ISCR Implementations on CNN for Improving Orientation Detection in Models","archived":false,"fork":false,"pushed_at":"2024-07-28T15:42:13.000Z","size":40317,"stargazers_count":1,"open_issues_count":0,"forks_count":0,"subscribers_count":1,"default_branch":"main","last_synced_at":"2025-02-23T04:16:36.930Z","etag":null,"topics":[],"latest_commit_sha":null,"homepage":null,"language":"Jupyter Notebook","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/YashBit.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","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}},"created_at":"2023-06-26T13:21:04.000Z","updated_at":"2024-07-28T15:42:19.000Z","dependencies_parsed_at":null,"dependency_job_id":"9a5dcfe9-b300-4165-8c44-39693b6b7918","html_url":"https://github.com/YashBit/ISCR_CNN","commit_stats":null,"previous_names":["yashbit/iscr_cnn"],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/YashBit/ISCR_CNN","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/YashBit%2FISCR_CNN","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/YashBit%2FISCR_CNN/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/YashBit%2FISCR_CNN/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/YashBit%2FISCR_CNN/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/YashBit","download_url":"https://codeload.github.com/YashBit/ISCR_CNN/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/YashBit%2FISCR_CNN/sbom","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":259549565,"owners_count":22875098,"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":[],"created_at":"2024-11-09T11:35:42.265Z","updated_at":"2025-06-23T08:34:38.935Z","avatar_url":"https://github.com/YashBit.png","language":"Jupyter Notebook","funding_links":[],"categories":[],"sub_categories":[],"readme":"# Project TODO List\n\n## W\u0026B Integration\n\n- [ ] Set up W\u0026B project for ISCR group: [ISCR W\u0026B Project](https://wandb.ai/iscr)\n\n## High-Level Overview\n\n### Images\n\n- **Compound/Continuous Set**\n    - [ ] Prepare 2000 images for each letter\n        - [ ] 1000 for continuous shapes\n        - [ ] 1000 for compound shapes\n    - **Total**: 52,000 images\n\n### First Blocker\n\n1. **How to Make Images Compound (Dotted Shapes)**\n   - [ ] Gilles to provide dataset for compound letters\n\n### Models\n\n1. **Model Parameters**\n    - [ ] Specify if ImageNet Pretrain is required (Y/N)\n    - [ ] Define finetuning details (Y/N, dataset, parameters)\n    - [ ] Specify mini datasets as needed for the image task\n\n2. **Results**\n    - [ ] Save model weights online\n    - [ ] Integrate with Weights and Biases for comprehensive tracking\n\n3. **Ablation Testing**\n    - [ ] Script for ablating layers (parameters: layer number)\n    - [ ] Measure performance metrics after ablating layers\n\n4. **Build Charts**\n    - [ ] Create performance charts similar to Paper 1\n        - [ ] Human Similarity (Relative Score)\n        - [ ] ...\n\n## Experiment 1: Compound vs Continuous Letters\n\n1. **Cognitive Science Concept Derived from Hypothesis**\n    - [ ] Define hypothesis related to orientation of letters\n    - [ ] Identify cognitive concept for experimentation\n\n2. **Human Experimental Suite / Data**\n    - [ ] Obtain experimental data from Davida (5.8/5.9)\n\n3. **Computational Experiment**\n    - [ ] Establish relation between DNNs and visual stream\n    - [ ] Formulate assumptions related to cognitive concept\n    - [ ] Prove cognitive concept using computational experiment\n\n4. **Psychological Representations to Computational Experiment**\n    - [ ] Establish connection between psychological and computational experiments\n\n5. **Theory for Experiment**\n    - [ ] Refine theory with Gilles\n    - [ ] Formulate theories based on human capability and model training\n\n6. **Experiment Details**\n    - [ ] Train models on compound and continuous exemplars\n    - [ ] Test models on compound/continuous exemplars\n    - [ ] Conduct ablative analysis on model layers\n\n## Engineering Methodology\n\n1. **Models Other Than DNNs**\n    - **Shallow Models:**\n        - [ ] Pixelwise\n        - [ ] GaborJet\n        - [ ] Histogram of Oriented Gradient\n        - [ ] Pyramid Histogram of Oriented Gradient\n        - [ ] Pyramid Histogram of Visual Words\n    - **HMAX Models:**\n        - [ ] HMAX 99’\n        - [ ] HMIN\n        - [ ] HMAX-PNAS\n\n2. **Deep Models**\n    - **Models:**\n        - [ ] GoogleLeNet\n        - [ ] VGG\n        - [ ] ResNet\n    - **Image Set:**\n        - [ ] Prepare 2000 images for each letter\n            - [ ] 1000 for continuous shapes\n            - [ ] 1000 for compound shapes\n        - **Total**: 52,000 images\n    - **Training Decisions:**\n        - [ ] Pretrained on ImageNet + finetuned on use case\n        - [ ] Fully trained only on use case, no ImageNet weights\n\n3. **Evaluation**\n    - [ ] Create bar charts of performance\n    - [ ] Conduct ablative testing for each model and layer\n\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fyashbit%2Fiscr_cnn","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fyashbit%2Fiscr_cnn","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fyashbit%2Fiscr_cnn/lists"}