This Website contains a comprehensive dataset for underwater change detection with thousands of handsegmented groundtruth images. Many of the special difficulties of the underwater environment are depicted in the videos, for example: marine snow, caustics or color attenuation. The moving objects in the videos are always fishes, which swim in swarms or separately.

The evaluation of the set is done in 100 frames per video and each groundtruth images contains three categories.

background = white (value: 255)
foreground  = black (value: 0)
unsure        = grey (value: 120)

All 5 videos are available here.

Some first results for three different approaches can be seen in the table below. The Algorithms were tested on the Marine Snow video of the Dataset.

 Algorithm True Negatives
 True Positives
False Negatives
 False Positives
 F1-Score MCC
3,209,105 1,918,995 0.8361 0.8487
MoSeg 194,489,1989,462,542 4,571,069910,791 0.77170.7754

 Algorithm  True Negatives
 True Positives
False Negatives
 False Positives
190,377,856 14,980,7372,606,529 1,468,4780.88020.8702 
190,281,34915,181,3352,405,931 1,564,9850.88430.8743 
 MoSeg178,455,951 11,185,284  2,570,997 12,892,596 0.57950.5912

M. Radolko, E. Gutzeit "Video Segmentation via a Gaussian Switch Background Model and Higher Order Markov Random Fields", VISAPP 2015 - Volume I : Proceedings of the International Conference on Computer Vision Theory and Applications. SciTePress, 2015, pp. 537-544

Z. Zivkovic. “Improved adaptive Gausian mixture model for background subtraction”, International Conference Pattern Recognition, UK, August, 2004, The code is very fast and performs also shadow detection. Number of Gausssian components is adapted per pixel.

Z.Zivkovic, F. van der Heijden. “Efficient Adaptive Density Estimation per Image Pixel for the Task of Background Subtraction”, Pattern Recognition Letters, vol. 27, no. 7, pages 773-780, 2006.


P. Ochs, J. Malik and T. Brox, "Segmentation of Moving Objects by Long Term Video Analysis," in IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 36, no. 6, pp. 1187-1200, June 2014.