How imondol41/thesisreport works
The repository is presented as a project report for traffic-object detection and distance estimation on RSUD20K, rather than as a source-code tree. The principal documented workflow runs from image input through YOLO11 detection, distance estimation, and annotated output; model training and evaluation are separate capabilities. Relationships follow the README’s architecture description. No implementation excerpts were available, so paths identify only the visible report entries, not verified code modules.
Main entry points: README.md Model training
Components 11
Data input
- RSUD20K images
README.md - Preprocessing
README.md
Detection
- YOLO11 detector
README.md - Detection post-processing
README.md
Distance estimation
- Distance estimation
README.md - Visualization and output
README.md - Annotated images
README.md
Model development
- Model training
README.md - Performance evaluation
README.md - COCO pretrained weights
README.md
Other components
- Dataset operator
Connections 11
- Dataset operator → RSUD20K images: provides images
- RSUD20K images → Preprocessing: supplies images
- Preprocessing → YOLO11 detector: provides resized images
- YOLO11 detector → Detection post-processing: produces detections
- Detection post-processing → Distance estimation: provides boxes and classes
- Distance estimation → Visualization and output: provides distances
- Visualization and output → Annotated images: saves images
- COCO pretrained weights → Model training: initializes
- RSUD20K images → Model training: trains on dataset
- Model training → YOLO11 detector: produces trained model
- Model training → Performance evaluation: evaluates model