Developed an algorithm for detecting people in distress using a swarm of drones aimed at aiding rescue and relief operations during natural calamities.Incorporated an algorithm to geocode images and relay it back to base station and developed an application interface to plot and show these images.Awarded the second position all over India in Technology category.
Developed a solution in Python to simulate LIDAR cloud point data given any track with multiple objects of various shapes at different locations, taking the LIDAR’s FoV into account.
My team secured 1st position in problem statement 2. The aim of the event was to develop a safety system which seamlessly integrates with PixHawk, a commercially popular autopilot for drones. The solution consisted of an internet connected Safety Unit that sits along with the autopilot and a website to monitor safety parameters.
Worked on an algorithm to stitch geo-tagged images received from a drone and make a map of the region of interest using computer vision. The team was declared winners and the software was delivered along with a technical presentation.
Developed a borescope based inspection device enabling the operator to inspect the quality of the cylinder bore and check for cracks/wear-tear by non-invasive methods. The platform lets the operator create a 3D plot of the inner surface of the bore, which can be saved for further analysis.I Worked on image processing and fault detection using deep learning for the prototype. The platform was developed in collaboration with Tata Motors and the prototype was demonstrated in the Tata Innovation Day, Taj Mahal Hotel, Bombay.
Proposed a complete setup to detect superficial defects on blue leather to classify it in terms of quality. The classification was based on the number of holes, lacerations, tick bite marks amongst others, on leather. Worked on developing the image processing algorithm for TATA Innoverse Challenge. The system was developed in Keras.