ShiftDrive

Due to advancement in technology getting a car is cheap, but maintaining it is very expensive. It is also difficult to get a good mechanic. The proposed automated system for dealing with this problem can classify the damaged vehicle and then predict the estimated cost for the damage repair. Convolution Neural Network (CNN) can be used for the understanding, detecting and analyzing various classes of the damage in the minor and major parts of the car. The damages can be of any type like bumper dent, door dent, glass shatter, head lamp broken, scratch and smash. Moreover, this system also allows customization of car by enabling the user to put stickers on their car and have a new look of their car before physical customization. Along with all these features, users can interact with each other via provided built-in social network.

ShiftDrive is android based application. It’s a development project. The scope of ShiftDrive is very broad since every person interacts with the cars directly or indirectly and thus can be a potential user of this application. The major users of this application will be the car owners and workshop workers as they will get the most benefits and can explore more opportunities out of it.

Project Scope

1. Detect the car using images and extract its Make and Model.
2. Detect all damaged parts.
3. Estimate repair costs.
4. Get live biddings from mechanics and vendors.
5. Get best Repair options around.

Model Details

Dataset - Train images 7000 - Test Images 220---- "Imagenet" is used for getting pre trained weights using Transfer Learning ----- "Softmax" Activation Function -----"Adam" optimizer -----"Categorical crossentropy" Loss Function.

Installation

$ git clone https://github.com/ultralytics/yolov5 $ cd yolov5 $ pip install -r requirements.txt

Inference

''' $ python detect.py --source 0 # webcam file.jpg # image file.mp4 # video path/ # directory path/*.jpg # glob 'https://youtu.be/NUsoVlDFqZg' # YouTube video 'rtsp://example.com/media.mp4' # RTSP, RTMP, HTTP stream '''

Environment (Tools)