(README.md written with copilot)
A comprehensive deep learning project for detecting obstacles in off-road environments using convolutional neural networks, developed as a Master's thesis.
This repository contains the complete implementation of a CNN-based object detection system designed to identify obstacles in challenging off-road terrain. The system leverages computer vision techniques to enable autonomous navigation in unstructured environments.
Note: The dataset images and trained model weights are not included due to their large file size.
The dataset was collected and manually annotated using the CVAT (Computer Vision Annotation Tool) software.
Reference: Boris Sekachev, Nikita Manovich, et al. (2020). Computer Vision Annotation Tool (CVAT) [Computer software]. Available at https://github.com/cvat-ai/cvat
- DP/ - Main codebase directory
- train_code/ - Model training implementations and scripts
- predict_code/ - Inference pipeline for obstacle detection
- val_code/ - Model validation and evaluation scripts
- track_code/ - Object tracking implementation
- utils/ - Utility functions and helper modules
- runs/detect/ - Detection results, model training and validation results
- val_results/ - Detailed validation metrics
- Language: Python
- Core Framework: Convolutional Neural Networks - YOLO Ultralytics (TODO: reference)
- Application: Off-road obstacle detection and object tracking
- End-to-end CNN architecture for object detection
- Robust performance in challenging off-road environments
- Integrated object tracking capabilities
- Comprehensive validation pipeline with detailed metrics
- Modular code structure for easy customization and extension