Saket Anand
Associate Professor
B-410, New Academic Block
IIIT-Delhi
Tel:+91-11-2690 7425

Current Students

PhD Students

M.Tech.(Research) Students (Thesis)

M.Tech. Students (Thesis)

B.Tech. Students (Thesis)

  • Navvrat Rao
  • Aajay A. Devaraj and Dhvanil Sheth
  • Atharv Goel
  • Mehar Khurana and Prakhar Gupta

Prospective Students

If you find the following topics interesting, please get in touch. We may be able to use your contributions to these projects.

  • Visual Analytics for Wildlife Monitoring

    • The goal of this project is to identify the species and/or individual in an image. This project is in collaboration with Wildlife Institute of India, and we work with camera trap images obtained from the field. We plan to apply techniques from robust statistics and machine learning to ensure reliable detection and identification of animals in cluttered backgounds. The ideal candidate should have very strong programming skills and a knack for probabilistic/statistical methods.

    • Species segregation in camera trap images (Classification using species hierarchies to handle morphologically similar species)
    • Individual re-identification of tigers and leopards
    • Active Learning for species segregation
    • Domain Adaptation for camera-trap based species segregation
    • Domain Adaptation for individual re-identification

    This project has the goal of deploying models at the Wildlife Institute of India (WII), where these models will be used for various camera-trap based analyses tasks including the All India Tiger Estimation (AITE). The 2018-19 edition of the AITE made it to the Guinness World Records for the "Largest camera-trap wildlife survey".

  • Vision based Perception for Autonomous Driving and Road Safety

    • The following problems are of interest to this project

    • Road Surface Segmentation
    • Traffic Light Detection and Tracking
    • Pedestrian Detection and Tracking
    • Trajectory forecasting
    • Multi-sensor Calibration and Active Learning equipped Annotation Tool
    • 3D LIDAR based localization and 3D mapping
    • 3D LIDAR based object detection, segmentation, and tracking
    • Testing of autonomous driving in simulation, virtual and augmented reality

    The objective is to develop robust algorithms for the aforementioned problems, which can run in near-real time. We plan to solve these problems using Machine Learning tools and are looking for strong B.Tech., M.Tech., and Ph.D. students who are familiar with libraries like OpenCV, Scikit-learn, PyTorch and TensorFlow.

  • Remote Sensing and Satellite Imagery for Agriculture

    • The following problems are of interest to this project

    • Super-resolution using cross-sensor signals
    • Pose estimation of Satellite sensors
    • Pixel-level segmentation for crop-type classification
    • Sensor model and geometry aware learning algorithms

    The objective is to develop efficient algorithms for performing inference using satellite imagery in a manner that can provide accurate farm-level predictions. The emphasis would be on small-scale farms in India and elsewhere. We are interested in accurate predictions of key cropping parameters like crop-type, moisture availability, sowing / harvesting dates, and soil parameters like moisture and nutrients. The skills required for this project are ML/DL frameworks like PyTorch, TensorFlow, etc. and a good knowledge of linear algebra and probability and statistics. Sound knowledge of geometric computer vision and signal processing is a big plus. B. Tech. and M. Tech. students are welcome for a thesis opportunity.

Past Students

Ph.D.

Research Assistants

Masters

B. Tech. (BTP)