Dev Khant

Machine Learning Intern

Aiborne•  July 2022 - November 2022

• Developed a custom Object Detection + OCR model for reading Digital Odometer. As simple OCR even cloud based couldn't give good accuracy. The mAP for the detection model was 96%. • Worked on Instance Segmentation and Object Detection using Pytorch. Goal here was to detect the dents, scratches, scrapes, cracks in windshield etc. present on cars and estimate the repairing cost. Loss Function used here was Focal Loss. • Integrated MLflow for model tracking and Data pipeline for Classification tasks. The classification model was trained for determining the view of the car. It included 30+ classes and the accuracy was 87%. • Worked on Detection Transformer(DeTR) for instance segmentation. Experimented on DINO which is an enhanced version of DeTR and the mAP improved by 4%. • Also worked on Video Analytics which showed for about how much time a particular class was in frame using a classification model and also showed the bluriness of frames.

Software Engineer

resoluteai.in•  April 2022 - July 2022

• Mainly worked on Computer Vision projects using Tensorflow. It included Object Detection, Backgound removal and knowing the color and brightness of an image. • Contributed in ongoing projects using YOLO for Object Detection and OpenCV for tracking and counting objects. This was used in manufacturing industries for counting bottles, towels and boxes. And mAP achieved here was 97.5%. During Inference test video ran at around 50 FPS. • Got hands-on experience of annotating real world industry data using LabelMe and LabelImg for the custom detection models. • Deployed 2 POCs using Streamlit and Flask. One project was to detect defects in fabrics and calculate the area covered by it. Another project was extracting table from bills or receipts using OCR. Pytesseract was used for OCR.