Projects

• Led the team to be ranked among the top 4 teams across campus in the Machine Learning Quiz winning a prize and qualifying to the next round

• Worked on Dual Path Networks for Object Localization achieving 88% accuracy on Flipkart custom e-commerce product dataset reaching level 3.

• Worked to improve the state-of-the-art PWCNet model for optical flow estimation using adversarial loss.

• Trained the model using a semi-supervised learning approach(1:3 split) to remove the potential bias in the neural network architecture.

• Implemented Semantic Segmentation using Generative Adversarial Networks and Conditional Random Fields on Indian Road Dataset.

Working on developing a model for converting an animal sketch to an animal image using pix2pix as a base model.

Working on the intersection of Deep Learning, Material Design and Property prediction.

• Trained Mask R-CNN and Hybrid Task Cascade models for Instance Segmentation on the Indian Driving Dataset. Used mixed-precision and multi-scale training for better feature understanding.

• Trained DeepLabv3+ on the Indian Driving Dataset for Semantic Segmentation. Used Gradient accumulation for optimizing batch size vs crop size trade-off.

• Achieved 16th place finish in ICCV's AutoNue 2019 challenge.

• Implemented a pipeline using OpenCV for detecting lane lines in images and videos of roads in Python for navigation in an autonomous vehicle.

• Used Canny edge detection to detect edges, followed by Hough transform to detect prominent lines in the Region of Interest.

• Used transformation to LAB colourspace to effectively handle low-light images.