How leonardocorreia08/observador works
OBSERVADOR is an early-stage computer-vision project: its implemented workflow loads configuration, acquires camera frames or dataset images, runs YOLO detection or tracking, and writes visual and structured results; a separate evaluator compares tracking output with MOT ground truth. The README also describes future camera-network, spatial, event, memory, and world-model capabilities, which are distinguished here from implemented code. Edges reflect shown calls, reads, or documented concepts; unsampled implementation details are not assumed.
Main entry points: config_loader.py Config loader · main.py Live runner · evaluate_tracking.py MOT evaluator · run_detection.py Image detection
Components 21
Live perception
- Live runner
main.py - Config loader
src/utils/config_loader.py - Camera manager
src/cameras/camera_manager.py - YOLO detector
src/detection/detector.py - Training validator
src/validation/training_validator.py
Detection and tracking
- COCO validation
experiments/exp_001_detection/validate_coco.py - Image detection
experiments/exp_002_detection/run_detection.py - Sequence tracking
experiments/exp_003_tracking/run_tracking.py - Detection outputs
results/exp_002 - Tracking outputs
results/exp_003
Counting and evaluation
- Line-count runner
scripts/track_count.py - Line counter
src/trafficvision/counting.py - MOT evaluator
experiments/exp_003_tracking/evaluate_tracking.py - Evaluation reports
results/exp_003/evaluation
Conceptual capabilities
- Camera network
- Spatial intelligence
- Event intelligence
- Memory system
- World model
Other components
- Operator
- Local camera
Connections 20
- Operator → Live runner: starts
main.py - Live runner → Config loader: loads config
main.py - Live runner → Camera manager: discovers and opens
main.py - Local camera → Camera manager: provides frames
src/cameras/camera_manager.py - Live runner → YOLO detector: runs detection
main.py - YOLO detector → Config loader: resolves settings
src/detection/detector.py - Live runner → Training validator: checks training
main.py - COCO validation → Config loader: loads settings
experiments/exp_001_detection/validate_coco.py - Image detection → Config loader: loads settings
experiments/exp_002_detection/run_detection.py - Image detection → Detection outputs: writes reports
experiments/exp_002_detection/run_detection.py - Sequence tracking → Config loader: loads settings
experiments/exp_003_tracking/run_tracking.py - Sequence tracking → Tracking outputs: writes tracks
experiments/exp_003_tracking/run_tracking.py - MOT evaluator → Config loader: loads settings
experiments/exp_003_tracking/evaluate_tracking.py - MOT evaluator → Tracking outputs: reads predictions
experiments/exp_003_tracking/evaluate_tracking.py - MOT evaluator → Evaluation reports: writes metrics
experiments/exp_003_tracking/evaluate_tracking.py - Line-count runner → Line counter: updates crossings
scripts/track_count.py - Camera network → Spatial intelligence: supports interpretation
README.md - Spatial intelligence → Event intelligence: informs analysis
README.md - Event intelligence → Memory system: records events
README.md - Memory system → World model: informs representation
README.md