Detection and Tracking of Objects outside the line of sight using Support Vector Machines
Non Line of Sight Imaging is a recurring challenge with a wide variety of important applications. The most commonly proposed approaches to date for Non Line of Sight Imaging depend on exploiting time-resolved measurements which requires expensive, specialized, ultrafast lasers and detectors that must be carefully calibrated.
This project makes the following contributions to address the issue:
1.Develop an approach that uses only a 2D camera and spot light source to perform the imaging
2.Generate a synthetic data set that is rendered based on the proposed method and use the data set to perform tracking using a Support Vector Machine.
Software used:
1.Rendering of the images was done using Physically Based Renderer(PBRT)
2.Support Vector Machine implementation is done using python
The code and dataset have not been uploaded but can be provided on request