# DU_MetroRail Solution

Description
Link
🚗🔍
Approach Summary
Our approach harnesses the power of the Co-DETR model 🤖 and the BadODD dataset 📊 for robust object detection on the intricate roads of Bangladesh, addressing unique challenges for autonomous navigation.
📄
IEEE Format Paper
[Link](https://drive.google.com/file/d/1iF3Q1592VmXjgrdZK8i0rxu5EOigeERc/view?usp=sharing)
🖥️
Approach Presentation
[Link](https://docs.google.com/presentation/d/1rWYaTH43CLrU6WRhWF6Mi_S2PPLOfhUXHy4hoRJPv2U/edit?usp=drivesdk)
📊
EDA Notebook
[Link](https://www.kaggle.com/code/abdullaharean/badodd-a-complete-eda)
🔄
Data Augmentation Notebook
[Link](https://www.kaggle.com/code/abdullaharean/badodd-data-augmentation)
🖼️
Image Processing Notebook
[Link](https://www.kaggle.com/code/istiaqahmedfahad/badodd-image-prerocessing)
🔄
Dataset Conversion Notebook
[Link](https://www.kaggle.com/code/abdullaharean/badodd-yolo2coco)
📓
Training Notebook
[Link](https://www.kaggle.com/sakibbai/du-metrorail-training-notebook)
🔍
Inference Notebook
[Link](https://www.kaggle.com/code/sakibbai/du-metrorail-inference-notebook)
💾
Model Weight
[Link](https://www.kaggle.com/datasets/abdullaharean/du-metrorail-model)