Skip to content

Depth Estimation for Embedded Devices using USB3.0 stereo camera

Notifications You must be signed in to change notification settings

s-eswar/Depth-Estimation-

Repository files navigation

Depth-Estimation for Embedded devices-

Project for depth estimation on embedded devices contianing code snippets using variations of different stereo matching algorithms: Block Matching(BM), Semi-Global Block Matching, Graph cuts, Iterated conditional mode , Belief propagation, Tree weighted passing. The code for comparison of the SM algorithms, available in this repository has been modified with openMP directives for parallel processing. However, the original code is also availble here. For comparison of stereo matching algorithms, the anchor paper is this.

Work

This thesis for this project can be referred for better understanding.

Obstacle detection -

For Obstacle detection with toycar, we used libuvc tool for streaming data(images/video) from the stereo camera.

Data -

The data has been picked up from middlebury site.

Relevant Material-

The relevant material regarding the cameras, algorithms, embedded devices and other tutorials is available here.