DLOSC

Citation Author(s):
Yu
Song
Xuzhao
Huang
Submitted by:
Xuzhao Huang
Last updated:
Mon, 07/08/2024 - 15:59
DOI:
10.21227/zdjh-4d76
Data Format:
*.bmp(zip) & *.png(zip)
License:
Creative Commons Attribution
36 Views
Categories:
Machine Learning
Keywords:
Deformable linear objects, DLOs Segmentation, Crossing types classification
0
0 ratings - Please login to submit your rating.
ACCESS DATASET CITE SHARE/EMBED

Abstract 

This dataset was created for Deformable Linear Objects(DLOs) segmentation and crossings classification under complex background.

RGB images, overlap maps, gradient maps are included for segmentation task. The quantity of DLOs range from one to three.

Upper and lower type of crossings are defined for classification task.

Instructions: 

Segmentation:

  • Original images: RGB images captured by Realsense camera with DLOs placed on four different backgrounds. The quantity of DLOs ranges from one to three.
  • Overlap maps: The extracted contour of DLOs. The overlap parts(i.e., crossings) of DLOs are in a different color with the contour.
  • Gradient maps: The extracted segment contour is a colored gradient according to its tracing route.
  • Image size: 640x320
  • Image formats: .bmp
  • Files: Single_DLO -- one rope under four complex backgrounds; Multi_DLOs -- two or three ropes under four complex backgrounds

Classification:

  • Image size: 50x50
  • Image format: .png
  • Categories: Upper and Lower

 

Dataset Files

LOGIN TO ACCESS DATASET FILES

 

QUESTIONS?