Aswathi M

Car Damage Detection and Analysis using Computer Vision - Collected and annotated approximately 20,000 client data using VGG annotation tool. Car body parts and damages were annotated separately for getting the location of damages. Trained the model with Mask RCNN using transfer learning. OCR for Reading Odometer and Chassis plate - Implemented a Deep learning model to detect and extract characters and number from a given License plate, Odometer or Chassis plate image using Faster RCNN and Tensorflow Object detection API. Smart Screen Insurance - Collected images of phone, built a classifier using MobileNet to determine if the uploaded image is a phone or non-phone. Used OpenCV techniques to determine whether the phone screen is cracked or not. Implemented CRAFT text detector using PyTorch to detect the text regions in the screenshots uploaded by the user and the detected text is cropped out and fed to Pytesseract for character recognition. Image Classifiers - Designed and developed different image classifiers like Binary classifier, Multi label classifier and Multiclass classifier for various client use cases using MobileNet and deployed them to Mobile devices using Tensorflow lite. License Plate Character Recognition using KNN Algorithm - Identify License plate numbers and characters. Implemented the model using KNN Algorithm and various OpenCV Techniques like thresholding, Morphological transformation, Image filters etc. Glass and Light Damage Analysis - Collected and annotated images of damaged windshields and lights, trained the model using YoloV4 with three classes say High damage, Medium damage and Low damage. Form Recognizer using Azure Form Recognizer - Annotated and trained custom form data to recognize and store meaningful data from Forms using Azure Form Recognizer, integrated it with Django and created API.