Computer Vision

About

What we do

Coding and Algorithms with help of Computer vision to detect Image Patterns and obstacles during the Flight

With an object, we do magic with

    Detection and localization


    We started with the problem of human detection in aerial imagery. Though the final goal is to be able to do that from a height range at which a UAV flies ( 200 to 400 ft. ), because of lack of a proper dataset at the time, we started with the problem of human detection in aerial imagery. Though the final goal is to be able to do that from a height range at which a UAV flies ( 200 to 400 ft. ), because of lack of a proper dataset at the time, we started off with using the INRIA person dataset. An SVM was trained using HOG features extracted from this dataset. Then, we picked up a random video from YouTube and applied this detection strategy. Here is the link to the result.

    Some Pictures related to the result:

    Classification


    Again, the final goal is to be able to detect and recognise letters in aerial imagery collected from a height of about 200 ft. to 400 ft. The dimension of letters being of about 8 ft. by 8 ft. To start off with the task we scaled down the whole problem and collected some data. Implementation of the detection and recognition strategy is underway. Here is the link to the result.

    Some Pictures related to the result: