{"id":25736929,"url":"https://github.com/phdenzel/deep-gesture","last_synced_at":"2026-04-11T02:34:20.827Z","repository":{"id":146779614,"uuid":"427069113","full_name":"phdenzel/deep-gesture","owner":"phdenzel","description":"An LSTM action recognition program that facilitates data collection for training and allows for real-time processing","archived":false,"fork":false,"pushed_at":"2022-10-19T16:20:20.000Z","size":670,"stargazers_count":0,"open_issues_count":1,"forks_count":1,"subscribers_count":2,"default_branch":"master","last_synced_at":"2025-06-07T20:43:50.280Z","etag":null,"topics":["keras","mediapipe","opencv-python","python3","tensorflow"],"latest_commit_sha":null,"homepage":"","language":"Jupyter Notebook","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"gpl-3.0","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/phdenzel.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":"2021-11-11T16:28:40.000Z","updated_at":"2021-11-11T20:53:11.000Z","dependencies_parsed_at":null,"dependency_job_id":"7d9d1c1e-d7a0-4546-a48a-8403f2d2d24c","html_url":"https://github.com/phdenzel/deep-gesture","commit_stats":null,"previous_names":[],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/phdenzel/deep-gesture","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/phdenzel%2Fdeep-gesture","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/phdenzel%2Fdeep-gesture/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/phdenzel%2Fdeep-gesture/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/phdenzel%2Fdeep-gesture/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/phdenzel","download_url":"https://codeload.github.com/phdenzel/deep-gesture/tar.gz/refs/heads/master","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/phdenzel%2Fdeep-gesture/sbom","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":260126689,"owners_count":22962674,"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":["keras","mediapipe","opencv-python","python3","tensorflow"],"created_at":"2025-02-26T06:31:48.089Z","updated_at":"2025-10-27T02:06:47.704Z","avatar_url":"https://github.com/phdenzel.png","language":"Jupyter Notebook","funding_links":[],"categories":[],"sub_categories":[],"readme":"\n\n# deep-gesture\n\nAn LSTM gesture-recognition neural net which can be trained to\ncategorize any number of given gestures.\n\nKey features:\n\n-   easy to train\n-   uses MediaPipe to generate a Holistic model for compact data collection\n-   a custom TensorFlow LSTM neural net\n\n![img](./imgs/example_ok.jpg)\n\n![img](./imgs/example_bad.jpg)\n\n\n## Requirements\n\n-   numpy\n-   scipy\n-   matplotlib\n-   opencv-python\n-   mediapipe\n-   sklearn\n-   tensorflow\n-   tensorflow-gpu\n\n\n## Install\n\nSoon, you'll simply be able to type `pip install deep_gesture`.\n\nTo install from source, you may type\n\n    pipenv install --dev\n    pipenv install -e .\n\n\n## Usage\n\nFor more information type\n\n    [pipenv run] deep_gesture -h\n\nFirst time `deep_gesture` is run, it creates a directory in your home\nfolder `~/.deep_gesture`. It will save the recorded data in\n`~/.deep_gesture/data` and TF models in `~/.deep_gesture/models`.  For\nproper training, it is recommended to record at least 20 sequences per\ngesture. Model training takes depending on the hardware 15-30 minutes.\nIf some problem occurs, it might be necessary to delete these\ndirectories, and start over.\n\nRun `deep_gesture` in collection mode (use webcam for data collection):\n\n    [pipenv run] deep_gesture --collect --device 0 --gestures hello --sequences 20 --length 30\n\nRun `deep_gesture` to train a model on collected data:\n\n    [pipenv run] deep_gesture --train --optimizer Adam --lr 0.0001 --epochs 1000 --batch-size 16\n\nRun `deep_gesture` in streaming mode (use webcam for real-time gesture recognition):\n\n    [pipenv run] deep_gesture --device 0\n\n(Not yet implemented) Run `deep_gesture` in file mode (use video file to categorize a gesture):\n\n    [pipenv run] deep_gesture --file example.mp4\n\n(Not yet implemented) Run `deep_gesture` in test-mode:\n\n    [pipenv run] deep_gesture -t\n\n(not yet implemented) or with `pytest`:\n\n    [pipenv run] pytest -v --cov=deep_gesture --cov-report=html\n\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fphdenzel%2Fdeep-gesture","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fphdenzel%2Fdeep-gesture","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fphdenzel%2Fdeep-gesture/lists"}