{"id":19480339,"url":"https://github.com/hasibzunair/unconventional-wisdom","last_synced_at":"2025-10-10T21:05:31.238Z","repository":{"id":118568777,"uuid":"136528487","full_name":"hasibzunair/unconventional-wisdom","owner":"hasibzunair","description":"[ICBSLP'2018] 6th place solution to Kaggle Bengali Handwritten Higit Recognition","archived":false,"fork":false,"pushed_at":"2019-08-02T00:07:52.000Z","size":23977,"stargazers_count":19,"open_issues_count":0,"forks_count":2,"subscribers_count":0,"default_branch":"master","last_synced_at":"2025-06-20T16:51:44.545Z","etag":null,"topics":["computer-vision","deep-learning","neural-network","transfer-learning"],"latest_commit_sha":null,"homepage":"https://www.researchgate.net/publication/326989744_Unconventional_Wisdom_A_New_Transfer_Learning_Approach_Applied_to_Bengali_Numeral_Classification","language":"Jupyter Notebook","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/hasibzunair.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}},"created_at":"2018-06-07T20:36:37.000Z","updated_at":"2025-05-20T13:55:47.000Z","dependencies_parsed_at":"2023-07-04T19:35:51.411Z","dependency_job_id":null,"html_url":"https://github.com/hasibzunair/unconventional-wisdom","commit_stats":null,"previous_names":[],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/hasibzunair/unconventional-wisdom","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/hasibzunair%2Funconventional-wisdom","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/hasibzunair%2Funconventional-wisdom/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/hasibzunair%2Funconventional-wisdom/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/hasibzunair%2Funconventional-wisdom/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/hasibzunair","download_url":"https://codeload.github.com/hasibzunair/unconventional-wisdom/tar.gz/refs/heads/master","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/hasibzunair%2Funconventional-wisdom/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":279005332,"owners_count":26083883,"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","status":"online","status_checked_at":"2025-10-10T02:00:06.843Z","response_time":62,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"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":["computer-vision","deep-learning","neural-network","transfer-learning"],"created_at":"2024-11-10T19:58:43.893Z","updated_at":"2025-10-10T21:05:31.198Z","avatar_url":"https://github.com/hasibzunair.png","language":"Jupyter Notebook","funding_links":[],"categories":["NLP Tools, Scripts and Utilities (also Projects)"],"sub_categories":["OCR/HTR"],"readme":"## What is it about?\n\nSixth place solution to kaggle bengali handwritten digit recognition. Achieved 97.606%. A VGG16 architecture pre-trained on imagenet used as a baseline with hyperparameter tuning, fine tuning intermediate layers, data augmentation and test time augmentation.\n\nThe unconventional approach led to surprising results which caught Jeremy Howard's attention which he ended up [tweeting about](https://twitter.com/jeremyphoward/status/1050427625011703808)! \n\n## Usage \n\nThis architecture is implemented in Python 3.6 and Keras using Tensorflow as backend.\n\n### Dependencies\n\nTested code using:\n\n*    Ubuntu 14.04\n*    Python 3.6\n\n### Directory Structure \u0026 Usage\n* `codes`: Contains codes to final submission\n* `others`: Contains helper codes and experimental notebooks. VERY MESSY!\n\n\n## Data set  \n* [Numta DB](https://www.kaggle.com/c/numta/data)\n\n## Cite\n\nIf you find this work useful in your research, please consider citing:\n```\n@inproceedings{zunair2018unconventional,\n  title={Unconventional Wisdom: A New Transfer Learning Approach Applied to Bengali Numeral Classification},\n  author={Zunair, Hasib and Mohammed, Nabeel and Momen, Sifat},\n  booktitle={2018 International Conference on Bangla Speech and Language Processing (ICBSLP)},\n  pages={1--6},\n  year={2018},\n  organization={IEEE}\n}\n```\nYou can also find it in IEEE Xplore Digital Library [here](https://ieeexplore.ieee.org/document/8554435)\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fhasibzunair%2Funconventional-wisdom","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fhasibzunair%2Funconventional-wisdom","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fhasibzunair%2Funconventional-wisdom/lists"}