Depth Video Segmentation Results and Datasets

Corresponding Author: Farzad Husain (shusain@iri.upc.edu)

On this page you can see the segmentation/tracking results using our algorithm. You can also download the associated datasets.
The page is organized as follows:
  1. Brief Explanation
  2. Scenario: Segmentation of a single Frame
  3. Scenario: Video segmentation with static camera
  4. Scenario: Video segmentation with moving camera
  5. Additional Results
  6. File Format


Brief Explanation
In the following videos you can see the segmentation results using only depth data. Each segment carries a unique label which is displayed on the centroid of the segment. The labels will be best visible if the video is set to higher resolution.


Scenario 1  The algorithm was run on a single frame repeatedly to achieve initial segmentation.

Figure 4: Video (Sensor -> Kinect camera) | Dataset download | Segmentation Result download

   


Figure 5: Video (Sensor -> LMS-Z210 by Reigl) | Dataset External link to webpage and dataset | Segmentation Result download




Table 1, ABW Dataset Video (Sensor -> ABW structured light camera) | Dataset: External link to webpage and dataset | Segmentation Result download

 
  

                                               
Figure 13: (Sensor -> ABW structured light camera) | Dataset External link to webpage and dataset | Segmentation Result download

AverageSegmentCoveringMetric_noise



Table 1, Perceptron Dataset: (Sensor -> Perceptron LADAR camera) | Dataset External link to webpage and dataset | Segmentation Result download



Figure 14: (Sensor -> Kinect camera) | Dataset External link to webpage and dataset | Segmentation Result : processed depths , raw depths




Cornell-RGBD-Dataset

External link to their webpage, where you can download their dataset.
From here you can also donwload our segmentation results.



Kinect-segmentation-Dataset


External link to their webpage, where you can download their dataset.
From here you can also donwload our segmentation results.



Scenario 2 : The camera pose was kept static while recording data.

Figure 6: Video (Sensor -> Kinect camera) | Dataset download | Segmentation Result download | Ground Truth download


Figure 7: Video (Sensor -> Kinect camera) | Dataset : download | Segmentation Result download | Ground Truth download
Moving object using robot arm + gripper


Figure 8: Video (Sensor -> Kinect camera) | Dataset download | Segmentation Result download | Ground Truth download


Figure 9: Video (Sensor -> Kinect camera) | Dataset download | Segmentation Result download


Figure 10: Video (Sensor -> Kinect camera) | Dataset download | Segmentation Result download



Scenario 3 : The camera was moving while the environment was static.

Figure 11: Video (Sensor -> Odetics LADAR camera | Dataset External link to webpage and dataset | Segmentation Result download




Figure 12: Video (Sensor -> TOF camera) | Dataset External link to webpage and dataset | Segmentation Result download | Ground Truth download



Additional Results

Video (Sensor -> Kinect camera) | Dataset download | Segmentation Result download

   




Weighted Score Unweighted Score
NYU_dataset_v2
53.55
26.01


File Format:
  The *mat files contain datasets which are provided by us. It will load the XYZ and RGB data for each frame in the form of a grid resulting in an array of size (l × w × 3 × n). Where l and w are the frame length and width respectively and n is the number of frames.

Copyright

The documents contained in these directories are included by the contributing authors as a means to ensure timely dissemination of scholarly and technical work on a non-commercial basis. Copyright and all rights therein are maintained by the authors or by other copyright holders, notwithstanding that they have offered their works here electronically. It is understood that all persons copying this information will adhere to the terms and constraints invoked by each author's copyright. These works may not be reposted without explicit permission of the copyright holder.