Weighted Unsupervised Learning for 3D Object Detection

Referenced paper : Weighted Unsupervised Learning for 3D Object Detection

3D Object Detection:

This paper introduces a novel weighted unsupervised learning for object detection using an RGB-D camera. This technique is feasible for detecting the moving objects in the noisy environments that are captured by an RGB-D camera. The main contribution of this paper is a real-time algorithm for detecting each object using weighted clustering as a separate cluster. In a preprocessing step, the algorithm calculates the pose 3D position X, Y, Z and RGB color of each data point and then it calculates each data point’s normal vector using the point’s neighbor. After preprocessing, our algorithm calculates k-weights for each data point; each weight indicates membership. Resulting in clustered objects of the scene.

Pipeline of 3D Object detection using RGB-D camera has two main parts: 1) Preprocessing including Mapping, Back-Projection, Normal Generating, Background removal and 2) Clustering including assigned initial weight, distance calculation,update weight and assign color, and finally visualization to illustrate the results.

Results:

Kinect color frame (RGB) with resolution of 1920 X 1080; b) Kinect depth frame with resolution of 512 X 424; c) Proposed method object detection using k= 15 clusters, and after 15 iterations.

a) Kinect color frame (RGB) with resolution of 1920 X 1080; b) Kinect depth frame with resolution of 512 X 424; c) Proposed method object detection using k= 7 clusters, and after 10 iterations. Memory consumption is 320 MB and framerate is 8.1±0.2FPS.

Results of segmenting scene objects using proposed algorithm;a) Segmentation of small duck;b) Segmentation anddetection of piece of red paper;c) Object detection of a box;d) Shows handy bag;e) Segmentation of box, the border of thebox has lower weight and it will be completed after several iteration;f) Representation of moving object, segmentation of aperson;g) Segmentation of basketball.

Citations

@article{Kowsari2016,
title = {Weighted Unsupervised Learning for 3D Object Detection},
journal = {International Journal of Advanced Computer Science and Applications}
doi = {10.14569/IJACSA.2016.070180},
url = {http://dx.doi.org/10.14569/IJACSA.2016.070180},
year = {2016},
publisher = {The Science and Information Organization},
volume = {7},
number = {1},
author = {Kamran Kowsari and Manal H. Alassaf},
}