Detectron2 for car damage detection using custom dataset

AI methods for Car Damage Detection with Mask-RCNN

A deep learning–based computer vision training pipeline for car damage detection using a Co-DETR learner enhanced with CBAM Attention, Hybrid Loss, and Albumentations. Trains on Colab to identify and localize car body defects such as scratches, dents, and rust. Includes end-to-end model training and quantitative evaluation.

The project has been executed in 2 methods one using Yolov5 for which u can see the demonstration in through the following link https://drive.google.com/file/d/1WTUw_j_NX_CqfcwwI7lozLnORTbWgHLp/view?usp=sharing and for the other approach using Detectron2 we have deployed it on hugging face

Computer vision model for automated vehicle damage detection and localisation, trained on 9,900 images using YOLOv5s. Developed during an internship at AddInn Group.

Car Damage Detection System using AI

Vehicle Damage Repair Estimation App Using Computer Vision Models

Car Damage Detection & Classification: Independent study with Blockchain Presence at the University of Zurich.

Android app that classifies car damage using MobileNetV2 + TFLite. Trained on the CarDD dataset with 83% accuracy across 6 damage categories.

Car damage detection - KNN repository

# Car Damage Detection using Detectron2 This project leverages **Detectron2**, a state-of-the-art object detection library, to detect car damages from images. The goal is to develop a model that can automatically identify and classify car damages, such as dents, scratches

An automated vehicle damage assessment tool powered by a YOLO26 instance segmentation pipeline. Identifies car parts, detects defects, and generates structured PDF reports with objective severity scoring.

CrashLens is a graduation project that applies computer vision and deep learning techniques to vehicle damage assessment. The system analyzes accident images to detect damage, classify severity levels, and estimate repair costs, supporting faster and more consistent preliminary evaluations. The project focuses on practical dataset experimentation,

Detect and localize vehicle damage in images using a custom YOLOv5 model to speed up insurance claims with accurate damage assessment.