How surya2324/object-detection works
This is a small real-time object-detection application. The workflow starts in model_data/main.py, which validates the configured inputs and creates a detector; model_data/Detector.py loads the model and class labels, captures frames, detects and filters objects, annotates the frames, and displays them. The entry point uses camera index 0; the sample videos are present in the tree but are not wired into the shown workflow. OpenCV behavior is inferred from the inspected calls.
Main entry points: Detector.py Video detector · main.py Application entry · frozen_inference_graph.pb Detection model · ssd_mobilenet_v3_large_coco_2020_01_14.pbtxt Model config
Components 9
Application entry
- Application entry
model_data/main.py
Detection runtime
- Video detector
model_data/Detector.py - OpenCV
Detection assets
- Detection model
model_data/frozen_inference_graph.pb - Model config
model_data/ssd_mobilenet_v3_large_coco_2020_01_14.pbtxt - Class labels
model_data/coco.names
Other components
- User
- Camera input
- Result window
Connections 14
- User → Application entry: starts
- Application entry → Detection model: validates path
- Application entry → Model config: validates path
- Application entry → Class labels: validates path
- Application entry → Video detector: creates
- Video detector → OpenCV: initializes DNN
- Video detector → Detection model: loads weights
- Video detector → Model config: loads config
- Video detector → Class labels: reads labels
- Video detector → OpenCV: captures frames
- Camera input → OpenCV: provides frames
- Video detector → OpenCV: detects and filters
- Video detector → OpenCV: annotates frames
- Video detector → Result window: shows frames