Implemented real-time end-to-end speech recognition system on hardware. Alongside, integrating a microphone, ADC and FPGA, and speakers through few pre-trained TCN/TDNN deep learning models. Link to Code
A step closer to L5 autonomy, we tried to solve a challenging problem that will help a vehicle make its own decisions by recognizing the traffic signals on the road, using BOSCH's dataset on Resnet models. Link to Code
Developed an Asset Performance Monitoring platform based on ML algorithms to calculate the maximum real-time reliability and efficiency of the process plant equipment such as motors, for the company GAIL. Link to Code
Building a Face-Detection Door Unlock system. Implemented various aspects such as face detection using pretrained haar-cascades, image processing using OpenCV. Learned to work with raspberry-pi devices. Link to Code
Outperformed the existing state of the art, Interacting Multiple Model (IMM) filter by deep-learning techniques, in-order to predict, associate, and smoothen out the objects' trajectories in radar vicinity, with a sound accuracy. Link to Code