How saranya-aluru/signova-dbms-project works
Signova is a prototype for real-time recognition of selected Indian Sign Language gestures. A user enables the webcam in the React interface; camera frames go to Flask, where MediaPipe landmarks are normalized and classified, and the result is presented in the interface. The documented backend also looks up sign details and records recognition history in MySQL; Kafka event publishing is an optional service capability. Data collection, extraction, and model-training modules support vocabulary development. Some backend stages and integrations are documented but their wiring was not fully sampled.
Main entry points: isl_dataset Gesture samples · app.py Flask API · CameraStage.jsx Camera capture · App.jsx Recognition UI
Components 26
Recognition Interface
- Camera capture
project/signova/src/CameraStage.jsx - Recognition UI
project/signova/src/App.jsx - React entry
project/signova/src/main.jsx - Sign library
project/signova/src/Signs.jsx
Recognition Backend
- Flask API
project/app.py - Hand detection
project/app.py - Landmark processing
project/app.py - Sign classification
project/app.py - Hand tracking
project/hand_tracking.py - Standalone recognizer
project/recognize.py
Data and Models
- Sample collection
project/collect_data.py - Two-hand collection
project/collect_two_hand.py - Dataset extraction
project/extract_dataset.py - Selected extraction
project/extract_selected.py - Model training
project/train_model.py - Expanded training
project/train_model_all.py - Two-hand training
project/train_two_hand.py - Gesture samples
project/isl_dataset - CSV samples
project/HELLO.csv
Persistence
- Database access
project/database_helper.py - MySQL database
- History viewer
project/view_history.py
Event Services
- Express event API
project/services/express-service/server.js - Kafka event API
project/services/kafka-service/server.js - Kafka broker
Other components
- Sign user
Connections 21
- Sign user → Camera capture: shows signs
README.md - React entry → Recognition UI: renders
project/signova/src/main.jsx - Recognition UI → Camera capture: renders
project/signova/src/App.jsx - Recognition UI → Sign library: renders
project/signova/src/App.jsx - Camera capture → Flask API: posts frames
project/signova/src/CameraStage.jsx - Flask API → Hand detection: detects hands
project/app.py - Hand detection → Landmark processing: extracts landmarks
project/app.py - Landmark processing → Sign classification: classifies features
project/app.py - Sign classification → Camera capture: returns recognition
- Flask API → Database access: looks up and saves
project/app.py - Database access → MySQL database: queries and records
project/database_helper.py - Model training → CSV samples: trains from
- Expanded training → Gesture samples: trains from
- Two-hand training → Gesture samples: trains from
- Sample collection → Gesture samples: collects samples
- Two-hand collection → Gesture samples: collects samples
- Dataset extraction → Gesture samples: extracts clips
- Selected extraction → Gesture samples: extracts clips
- Flask API → Kafka broker: publishes events
README.md - Express event API → Kafka broker: publishes events
project/services/express-service/server.js - Kafka event API → Kafka broker: publishes events
project/services/kafka-service/server.js