{"id":23406977,"url":"https://github.com/andystmc/hand2num","last_synced_at":"2026-04-28T08:04:26.477Z","repository":{"id":265226961,"uuid":"895515024","full_name":"AndysTMC/Hand2Num","owner":"AndysTMC","description":"A deep learning model that converts hand gestures into numerical values (1-5) using Convolutional Neural Networks for efficient and accurate recognition.","archived":false,"fork":false,"pushed_at":"2024-11-28T18:18:57.000Z","size":95886,"stargazers_count":0,"open_issues_count":0,"forks_count":0,"subscribers_count":1,"default_branch":"main","last_synced_at":"2025-04-09T00:07:21.613Z","etag":null,"topics":["classification","cnn","computer-vision","deep-learning","hand-gestures","machine-learning","media-pipe","numbers","opencv","real-time-prediction"],"latest_commit_sha":null,"homepage":"","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/AndysTMC.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}},"created_at":"2024-11-28T10:58:09.000Z","updated_at":"2024-11-28T18:19:00.000Z","dependencies_parsed_at":"2025-02-14T18:57:58.096Z","dependency_job_id":null,"html_url":"https://github.com/AndysTMC/Hand2Num","commit_stats":null,"previous_names":["andystmc/hand2num"],"tags_count":0,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/AndysTMC%2FHand2Num","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/AndysTMC%2FHand2Num/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/AndysTMC%2FHand2Num/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/AndysTMC%2FHand2Num/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/AndysTMC","download_url":"https://codeload.github.com/AndysTMC/Hand2Num/tar.gz/refs/heads/main","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":247947859,"owners_count":21023066,"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":["classification","cnn","computer-vision","deep-learning","hand-gestures","machine-learning","media-pipe","numbers","opencv","real-time-prediction"],"created_at":"2024-12-22T14:16:39.994Z","updated_at":"2026-04-28T08:04:21.437Z","avatar_url":"https://github.com/AndysTMC.png","language":"Jupyter Notebook","funding_links":[],"categories":[],"sub_categories":[],"readme":"# Hand2Num\n\n## Project Overview\n\nThis project is a real-time hand gesture recognition system that uses computer vision and deep learning technologies to classify hand gestures from webcam input. The system leverages MediaPipe for hand landmark detection and a custom Convolutional Neural Network (CNN) for gesture classification.\n![Theme_Image_Transparent](https://github.com/user-attachments/assets/a34aa079-684e-4368-8e70-8a9d87b1cdcd)\n\n\n## Key Features\n\n- Real-time hand gesture recognition\n- Uses MediaPipe for hand landmark detection\n- Custom CNN model for gesture classification\n- Developed and trained on Google Colab\n- Supports multiple gesture categories\n\n### Key Training Environment Features\n\n- Direct Google Drive file access\n- Compressed image dataset handling\n- Automated model training and checkpointing\n- GPU/TPU acceleration for faster computations\n\n\n## Technologies Used\n\n- **Development Platform**: Google Colab\n- **Hardware Acceleration**: TPU\n- **Computer Vision**: OpenCV (cv2)\n- **Hand Tracking**: MediaPipe\n- **Deep Learning**: TensorFlow/Keras\n- **Programming Language**: Python\n\n\n## Project Structure\n\n### 1. Data Generation (`generate.py`)\n- Captures hand landmark images using webcam\n- Processes and saves landmark images for training\n- Supports different hand configurations (left/right, normal/flipped)\n#### Preprocessed Images\n##### One-Left-Normal | One-Right Normal | One-Left Flipped | One-Right Flipped\n\u003cimg src=\"https://github.com/user-attachments/assets/cfb9ef40-a4b5-436c-8aa6-9b6b50a50e81\" alt=\"Image 1\" width=\"200\"/\u003e\n\u003cimg src=\"https://github.com/user-attachments/assets/3fd52b95-3d00-4b2a-82f1-70b1edd7d644\" alt=\"Image 1\" width=\"200\"/\u003e\n\u003cimg src=\"https://github.com/user-attachments/assets/f1ebfa43-1f92-4649-acbc-a6210d230cb8\" alt=\"Image 1\" width=\"200\"/\u003e\n\u003cimg src=\"https://github.com/user-attachments/assets/fafc9dd2-06ce-4e55-bee0-12efd94e8bea\" alt=\"Image 1\" width=\"200\"/\u003e\n\n##### Two-Left-Normal | Two-Right Normal | Two-Left Flipped | Two-Right Flipped\n\u003cimg src=\"https://github.com/user-attachments/assets/cca75418-486f-4ee2-b117-c7943655ed18\" alt=\"Image 1\" width=\"200\"/\u003e\n\u003cimg src=\"https://github.com/user-attachments/assets/05156f92-fed2-4260-90bf-fe365ceeac59\" alt=\"Image 1\" width=\"200\"/\u003e\n\u003cimg src=\"https://github.com/user-attachments/assets/a8594e12-847d-41a3-bc83-255158a2503f\" alt=\"Image 1\" width=\"200\"/\u003e\n\u003cimg src=\"https://github.com/user-attachments/assets/8f1e0b95-b3b3-43c6-ac98-7e7ef3390d0d\" alt=\"Image 1\" width=\"200\"/\u003e\n\n##### Three-Left-Normal | Three-Right Normal | Three-Left Flipped | Three-Right Flipped\n\u003cimg src=\"https://github.com/user-attachments/assets/8e687524-cd90-4252-9a9e-f414b783c1f5\" alt=\"Image 1\" width=\"200\"/\u003e\n\u003cimg src=\"https://github.com/user-attachments/assets/fffe768d-7a30-4b4e-97fb-072101831cd2\" alt=\"Image 1\" width=\"200\"/\u003e\n\u003cimg src=\"https://github.com/user-attachments/assets/a095a547-ffc4-4dfd-a462-267124bd2015\" alt=\"Image 1\" width=\"200\"/\u003e\n\u003cimg src=\"https://github.com/user-attachments/assets/3e137109-d1d4-4551-b312-29210f742144\" alt=\"Image 1\" width=\"200\"/\u003e\n\n##### Four-Left-Normal | Four-Right Normal | Four-Left Flipped | Four-Right Flipped\n\u003cimg src=\"https://github.com/user-attachments/assets/049f96fe-3da6-4a1e-beab-12e42d08662e\" alt=\"Image 1\" width=\"200\"/\u003e\n\u003cimg src=\"https://github.com/user-attachments/assets/f577d18f-ce00-4d3f-aee9-b3e80a3f9c54\" alt=\"Image 1\" width=\"200\"/\u003e\n\u003cimg src=\"https://github.com/user-attachments/assets/0590b82e-c2c3-4fdd-830b-f00534432982\" alt=\"Image 1\" width=\"200\"/\u003e\n\u003cimg src=\"https://github.com/user-attachments/assets/be261bc3-39f4-4128-9795-c8b5f99153ca\" alt=\"Image 1\" width=\"200\"/\u003e\n\n##### Five-Left-Normal | Five-Right Normal | Five-Left Flipped | Five-Right Flipped\n\u003cimg src=\"https://github.com/user-attachments/assets/a7decded-e83a-4fe3-84bd-0a074326ea28\" alt=\"Image 1\" width=\"200\"/\u003e\n\u003cimg src=\"https://github.com/user-attachments/assets/f7a30dbc-905b-4f69-a576-491b38de19fb\" alt=\"Image 1\" width=\"200\"/\u003e\n\u003cimg src=\"https://github.com/user-attachments/assets/25cff253-09c3-4469-9192-5f6b21cec02f\" alt=\"Image 1\" width=\"200\"/\u003e\n\u003cimg src=\"https://github.com/user-attachments/assets/b89d1bf3-7b02-4588-812a-f44467993050\" alt=\"Image 1\" width=\"200\"/\u003e\n\n\n### 2. Model Training (`Project_HGR.ipynb`)\n- Prepares and preprocesses image dataset\n- Builds a Convolutional Neural Network (CNN)\n- Trains and validates the gesture recognition model\n- Saves the best performing model\n### Model Architecture\n\u003cimg src=\"https://github.com/user-attachments/assets/6f82f1ee-f4f0-4d5b-b349-76d57b817959\" alt=\"Image 1\" width=\"900\"/\u003e\n\n### Few Testing Results\n\u003cimg src=\"https://github.com/user-attachments/assets/96399dd8-9fb6-4bdd-96c8-b7bb46fc11d4\" alt=\"Image 1\" width=\"900\"/\u003e\n\n### 3. Live Classification (`live_cam_test.py`)\n- Loads pre-trained model\n- Processes real-time webcam input\n- Performs hand gesture recognition\n- Displays prediction results\n### Some Real-time Testing Results\n\u003cimg src=\"https://github.com/user-attachments/assets/996dbc5b-e6cb-405f-a839-8947162bbe01\" alt=\"Image 1\" width=\"300\"/\u003e\n\u003cimg src=\"https://github.com/user-attachments/assets/ade4e0e9-375f-47e3-bff5-9a9e593b02ab\" alt=\"Image 2\" width=\"300\"/\u003e\n\u003cimg src=\"https://github.com/user-attachments/assets/6e78e9d6-b8ab-4855-a75c-981077c975f4\" alt=\"Image 3\" width=\"300\"/\u003e\n\u003cimg src=\"https://github.com/user-attachments/assets/b6dac283-8f1a-4273-b441-d30f05bfa6be\" alt=\"Image 4\" width=\"300\"/\u003e\n\u003cimg src=\"https://github.com/user-attachments/assets/7d1ebde8-cc89-4522-bde5-c216e18f9112\" alt=\"Image 5\" width=\"300\"/\u003e\n\n\n## Setup and Reproduction\n\n### Prerequisites\n- Google Account\n- Google Colab access\n- Prepared image dataset\n\n### Steps to Reproduce\n1. Open Google Colab\n2. Create new notebook\n3. Upload or link to required Python scripts\n4. Mount Google Drive\n5. Upload compressed image dataset\n6. Run training notebook (Project_HGR.ipynb)\n\n\n## Model Deployment\n\nAfter training in Colab:\n- Download the best performing model\n- Use `live_cam_test.py` for real-time gesture recognition\n- Ensure all dependencies are installed locally\n\n## Potential Improvements\n- Increase training dataset diversity\n- Implement data augmentation\n- Experiment with model architectures\n- Add more gesture categories\n\n## Limitations\n- Requires good lighting conditions\n- Performance depends on training data quality\n- Currently supports a limited number of gesture categories\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fandystmc%2Fhand2num","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fandystmc%2Fhand2num","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fandystmc%2Fhand2num/lists"}