Deep learning model that transcribes ancient tamil inscriptons from 10th century to modern tamil letters. It also involved using openCV to collect and build a dataset. This project was completed for Nithilam - a tamil research club.

A dynamic QC system was built using a 4k 60fps camera in order to precisely measure the dimensions of Brass bracket with very high precision (30 microns). The input was taken from a moving conveyor and was processed using various image processing techniques in opencv.

An AI based project that de-smokes real time smoky video frames and generates clear output in order to aid fire fighters in indoor fire hazards. It involved building a TF deep learning model that generates the gray scaled edges of the surrounding aiding in vision.

In this project, I have developed an evolutionary algorithm for feature selection in Machine Learning. Here, the individuals with the best possible features are allowed to produce further generations to select the best features from the data.

A Video Classification model was trained on 300 videos to classify original and deep fake videos. The faces in the videos were cut out by using mediapipe and 3d convolution layers were used to extract the temporal information from the videos.

In this project the x-ray images of lungs are classified into 3 categories viz, covid positive, covid nagative and other non-covid diseases. Vgg-16 based model has been trained for this project using tensorflow. The UI for this project wwas developed using streamlit