How zhaoj9014/anti-uav works
Anti-UAV is a computer-vision research repository for detecting and tracking UAVs in RGB and thermal video, with training and benchmark tooling and several PyTorch/Jittor implementations. A visible runtime path receives frames in the camera-link application, invokes detection and tracking, and can send target coordinates over UDP. The README also describes training and evaluation as product capabilities. The tree contains multiple independent implementations; unsampled subsystems are represented only where the README or source index supports their role, and the graph does not imply they form one pipeline.
Main entry points: Detectoruav.py UAV Detector · detect_tracking.py Camera Link · experimental.py YOLO Layers · tracker.py UAV Tracker
Components 19
Live Detection
- Camera Link
Codes/detect_tracking.py
Detection Models
- UAV Detector
Codes/detect_wrapper/Detectoruav.py - YOLO Layers
Codes/detect_wrapper/models/experimental.py - YOLO Components
Codes/detect_wrapper/models/common.py - Image Loading
Codes/detect_wrapper/utils/datasets.py - Detection Utilities
Codes/detect_wrapper/utils/general.py - Device Utilities
Codes/detect_wrapper/utils/torch_utils.py - Jittor Detector
anti_uav_jittor/dectect/dect_predict.py - Region Proposals
anti_uav_jittor/dectect/model/rpn.py - SiamDT Inference
anti_uav_jittor/anti_uav410_jit/trackers/SiamDT/libs/swintransformer/tools/deployment/mmdet_handler.py - Detector Inference API
anti_uav_jittor/anti_uav410_jit/trackers/SiamDT/libs/swintransformer/mmdet/apis/inference.py
Tracking
- UAV Tracker
Codes/tracking_wrapper/dronetracker/trackinguav/evaluation/tracker.py - FHOG Features
Codes/tracking_wrapper/drtracker/fhog.py
Training and Evaluation
- Detector Training
Codes/detect_wrapper/train_drone.py - Anti-UAV Dataset
anti_uav_jittor/anti_uav410_jit/datasets/antiuav410.py - Tracking Evaluation
anti_uav_jittor/anti_uav410_jit/Evaluation_for_ALL.py - Tracking Metric
Other components
- Video Source
- Coordinate Receiver
Connections 15
- Video Source → Camera Link: sends frames
Codes/detect_tracking.py - Camera Link → UAV Detector: requests detection
Codes/detect_tracking.py - Camera Link → UAV Tracker: uses tracker
Codes/detect_tracking.py - Camera Link → Coordinate Receiver: sends coordinates
Codes/detect_tracking.py - UAV Detector → YOLO Layers: loads models
Codes/detect_wrapper/Detectoruav.py - UAV Detector → Image Loading: loads images
Codes/detect_wrapper/Detectoruav.py - UAV Detector → Detection Utilities: uses utilities
Codes/detect_wrapper/Detectoruav.py - UAV Detector → Device Utilities: selects device
Codes/detect_wrapper/Detectoruav.py - YOLO Layers → YOLO Components: uses layers
Codes/detect_wrapper/models/experimental.py - UAV Tracker → FHOG Features: uses features
Codes/tracking_wrapper/dronetracker/trackinguav/evaluation/tracker.py - SiamDT Inference → Detector Inference API: runs inference
anti_uav_jittor/anti_uav410_jit/trackers/SiamDT/libs/swintransformer/tools/deployment/mmdet_handler.py - Jittor Detector → Region Proposals: constructs detector
anti_uav_jittor/dectect/dect_predict.py - Detector Training → YOLO Layers: initializes model
Codes/detect_wrapper/train_drone.py - Anti-UAV Dataset → Tracking Evaluation: evaluated by
- Tracking Evaluation → Tracking Metric: computes scores