[visionlist] Call for participation:Off road image segmentation challenge (ML competition)

[matsuzawa.toyoki.qs at cs.atla.mod.go.jp]

+++++++++++++++++++++++++++++++++++++++++++++++++
Off road image segmentation challenge (Machine Learning competition)
+++++++++++++++++++++++++++++++++++++++++++++++++

We are conducting the research on image based autonomous driving, and think
that
it is important to establish the image segmentation of the movable ground
plane and the obstacles with high accuracy and speed using machine learning,
especially deep learning approach.
Past few years, we have focused on the data acquisition of real off-road
vehicle with image sensor,  etc, and built the data set for image
segmentation.  
Now, using the part of the data set, we are pleased to announce the "Off
road Image segmentation challenge", some kind of machine learning
competition, to determine the segmentation accuracy and inference speed of
the proposed prediction model of the participants.

The participants will be asked to partition the images into the multiple
segment at the pixel level, such as the 19 categories of "road", "dirt road"
, "other obstacle", and etc.

The data set contains image data , approx. 3000 annotated data(19
categories),
approx. 2000 annotated data (2 categories), and would be used for your
research, and publications.

The competition is open for everyone in the world, without any restriction. 
The participation can be done with completely anonymous, and free of charge.

If you are interested, please access the following for the detail, and
download the data set;
https://signate.jp/competitions/101

#### Judging criteria ####

1.Segmentation Accuracy division
 mIOU of top three categories

2.Inference speed division
 Computation speed of image segmentation provided the mIOU is greater than
0.75

Participation is free of charge, and award will be presented to the top 3
contributors for each divisions.

#### Important date ####

Competition open: 9th December, 2020
Submission clos; 11th February, 2021

Best regards
Toyoki Matsuzawa, Ph.D.
Head of Intelligent Systems Research Section, ADTeC, ATLA
Tokyo Japan.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 7221 bytes
Desc: not available
URL: <http://visionscience.com/pipermail/visionlist_visionscience.com/attachments/20201224/4f2a7d6f/attachment.p7s>