How remyc0des/gestify works
Gestify is a webcam-based hand-gesture controller for Spotify, with a separate data-preparation and model-training workflow. The main runtime captures frames, recognizes gestures, visualizes landmarks and confidence, applies a cooldown, and dispatches Spotify commands. The repository also contains landmark extraction and a training script. The README documents the runtime behavior; source excerpts show the recognition, command, extraction, and training implementations. The extraction script references video files under a path not present in the supplied tree, so that input is not represented as a repository node.
Main entry points: main.py Frame Capture · extract_landmarks.py Landmark Extraction · modelbuild.py ResNet Training · Annot_List.txt Gesture Annotations
Components 12
Live Gesture Control
- Frame Capture
main.py - Gesture Recognition
main.py - Landmark Display
main.py - Cooldown Gate
main.py - Gesture Commands
main.py
Gesture Data
- Gesture Annotations
Annot_List.txt - Landmark Extraction
extract_landmarks.py
Model Training
- IPN Gesture Clips
IPN_dataset - ResNet Training
modelbuild.py
Other components
- User
- Webcam
- Spotify
Connections 11
- User → Frame Capture: starts
- Webcam → Frame Capture: provides frames
main.py - Frame Capture → Gesture Recognition: submits frames
main.py - Gesture Recognition → Landmark Display: provides results
main.py - Landmark Display → Cooldown Gate: checks gesture
main.py - Cooldown Gate → Gesture Commands: permits command
main.py - Gesture Commands → Spotify: sends AppleScript
main.py - User → Landmark Extraction: runs extraction
- Gesture Annotations → Landmark Extraction: labels videos
extract_landmarks.py - User → ResNet Training: runs training
- IPN Gesture Clips → ResNet Training: loads images
modelbuild.py