Skip to content

TilakD/Car-Damage-Analysis-App

Repository files navigation

Car Damage Detective

Assessing Car Damage with Convolutional Neural Networks

A proof of concept to use computer vision and deep learning to check whether a car is damaged or not and if damaged check severity and location. Trained a pipeline of convolutional neural networks using transfer learning on DenseNet-201 with Keras and Tensorflow to classify damage. Deployment is done using a web app with Flask, dockers and tensorflow serving.

Model Workflow: Alt text

Explainability: Alt text

Deployment: Alt text

Credits for the concept - Ting Neo's car-damage-detective

About

Assessing external car damage, i.e., severity and location using Deep Learning and deploy it using flask and tensorflow serving.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages