Implemented algorithm for product recommendation system

Implemented fish length detection system. This system locate the fish from the given image and then predict it's length based on the scale present in the image. This complex system was built for a fish catching organizer company. It automate the fish length calculation manual process and helped client in saving time and money.

Created an image classification model using deepl learning (keras) and deploy it as Rest API using Flask framework.

Created and deployed an algorithm to automate teh product categorization process for Lazada- biggest ecommerce sitre in Singapore.

Implemented Python 3.x scripts to extract, transform and load merchants data for one of the NTUC project. These scripts can perform ETL for 26 merchants avaialble in Singapore. This data helps NTUC in their consumer reward programme.

Extracted brand level details from shelf images. Images contains various wine brands bottles. Extracted information contains brand name, number of bottles, price. Implemented the object detection and recognition system using YOLOv3, keras API with TensorFlow backend, Python 3.7, OpenCV. Deployed the solution as Rest API using Flask.

Implemented a real time brand & price detection system for a winery distributor client.

This automated system helps to analyze their and competitors brands.

Sales user just need to upload or take a picture from his ipad app and on real time this system shows the brand names, no. of bottles, prices and it's segment.

The complete deep learning based system I have implemented using tiny YOLOv3, TensorFlow, OpenCV and Python.

Django based web application for digital signature functionality where a user can upload pdf or doc format document, drag and drop name/date/signature fields anywhere in the uploaded document, add single/multiple recipients and submit. Recipients get an email notification with a link to sign the document. On clicking the link the document is opened in a browser tab, recipients can add name/date/place signature and submit.  Sender gets the notification as soon as the recipient ha assigned the document and then he can view/download/print the signed document. The uploaded doc and signature are encrypted-decrypted using cryptography and digital signature protocols.

Detecting and Recognizing 400+ liquor brands on real-time video recording or recorded video file. This model has integrated with the iPhone app for one of my clients. This model is implemented using Python 3, YOLOv4-tiny and OpenCV.