1. A machine learning approach to expedite the FEM Analysis
In this project I with other two project members developed the ML-model for the prediction of Stress and deformation on a 3-D cantilever beam. This will accelerate the product development workflow by dramatically reducing the speed of FEM analysis.
Objective:
Accelerate the product development workflow by dramatically reducing the need for prototyping or iterative testing.Prediction of Stress and deflection for a given model and load.
Method:
FEMMachine LearningAnsys SoftwareAutomation (Python)
2. Structure Fault detection using LIDAR data
(IIT Delhi)
In this project, We are trying to leverage Machine Learning on 3-D Scan of a monument in Delhi (Dadi poti's tomb). Libraries like pytorch 3D (by facebook) and other LAS pacakages available in python are being used.
3. Crowd Tracking and Crowd parameter prediction in Kumbh mela using OpenCV and Optical flow
(IISC Bangalore)
In this project, We have optimised unsupervised Crowd tracking algorithms to track pedestrians, VGG-16 based models have been used to predict crowd density. Parameters from crowd movement can help prevent stampedes and improve event management capacties of the administration in Super-events like Kumbh mela.
Prakhar 