Introducing 

Prezi AI.

Your new presentation assistant.

Refine, enhance, and tailor your content, source relevant images, and edit visuals quicker than ever before.

Loading…
Transcript

INTRODUCTION

The project “Automobile Component Identification and collision damage analysis” aims to develop an advanced system that employs cutting-edge image processing and artificial intelligence techniques to accurately identify and categorize damages incurred by vehicles. Automobile damage analysis plays a vital role in various organizations, offering multifaceted benefits tailored to their specific needs.

For instance an insurance company can leverage this sophisticated model to effectively assess the type and precise location of damage based on the input image provided.

The model developed here is trained using the YOLO algorithm. YOLO, short for "You Only Look Once” is a state-of-the-art, real-time object detection system used in computer vision and deep learning.

Automobile component and collision detection

CURRENT STATUS

Give an update about various tasks within the project

DONE

WORKING

Listed initiative

Listed initiative

Listed initiative

TO DO

About dataset

IN PROGRESS

  • The dataset used here is from kaggle datasets provided by “Human In The Loop” titled “Car Parts and Car Damages” used for car part and car damage analysis AI systems for automated claims processing.

  • Total of 1812 images, fully annotated with polygons.

  • car parts=998 images

  • car damages=814

  • The total number of polygons in the dataset is 24,851.

  • annotations have been saved in .json format and the annotations are of polygon type but for the model described in this project the annotations used are of the yolo format (.txt) and is of rectangle type.

  • Car parts and damage dataset link- https://www.kaggle.com/datasets/humansintheloop/car-parts-and-car-damages?select=Car+damages+dataset

NEXT STEPS

Draw a conclusion and define areas for future efforts

YOLO

Yolo methodology

YOLO (You Only Look Once) is an object detection algorithm that rapidly and accurately identifies objects in images or videos, achieving real-time performance by simultaneously predicting bounding boxes and class probabilities in a single pass through the neural network.

Learn more about creating dynamic, engaging presentations with Prezi