The MVSeg dataset is available now!

-------

# Dataset Brief

MVSeg dataset comprises 738 calibrated RGB and thermal (RGB-T) videos, accompanied by 3,545 fine-grained pixel-level semantic annotations of 26 categories. It covers a wide range of challenging urban scenes in both daytime and nighttime.

-------

# Dataset Download

Please download the MVSeg dataset from the following Google Drive link: " https://drive.google.com/file/d/1duCXz5aUuvZG7HBnydHLL61k4UBT2Wu_/view?usp=sharing ". Kindly note that certain essential information, such as your name, affiliation, and purpose, is required. Once your request is approved, you will be granted access to the dataset.

-------

# Dataset Structure

The MVSeg is structured as follows:

MVSeg_Dataset/
|- train.txt
|- val.txt
|- test.txt
|- test_day.txt
|- test_night.txt
| ...
|- data/
|  |- video1/
|  |  |- visible/
|  |  |  |- 00001v.jpg
|  |  |  |- 00002v.jpg
|  |  |- infrared/
|  |  |  |- 00001i.jpg
|  |  |  |- 00002i.jpg
|  |  |- label/
|  |  |  |- 00001l.png
|  |
|  |- video2/
|  |  |- visible/
...

-------

# Dataset ColorMap

Here is the reference for MVSeg dataset color visualization.

[
        (0, 0, 0),          # 0:    background(unlabeled)
        (0, 0, 142),        # 1:    Car
        (0, 60, 100),       # 2:    Bus
        (0, 0, 230),        # 3:    Motorcycle
        (119, 11, 32),      # 4:    Bicycle
        (255, 0, 0),        # 5:    Pedestrian
        (0, 139, 139),      # 6:    Motorcyclist
        (255, 165, 150),    # 7:    Bicyclist
        (192, 64, 0),       # 8:    Cart
        (211, 211, 211),    # 9:    Bench
        (100, 33, 128),     # 10:   Umbrella
        (117, 79, 86),      # 11:   Box
        (153, 153, 153),    # 12:   Pole
        (190, 122, 222),    # 13:   Street_lamp
        (250, 170, 30),     # 14:   Traffic_light
        (220, 220, 0),      # 15:   Traffic_sign
        (222, 142, 35),     # 16:   Car_stop
        (205, 155, 155),    # 17:   Color_cone
        (70, 130, 180),     # 18:   Sky
        (128, 64, 128),     # 19:   Road
        (244, 35, 232),     # 20:   Sidewalk
        (0, 0, 70),         # 21:   Curb
        (107, 142, 35),     # 22:   Vegetation
        (152, 251, 152),    # 23:   Terrain
        (70, 70, 70),       # 24:   Building
        (110, 80, 100)      # 25:   Ground
]

-------

# Note

Our MVSeg dataset is under the MIT License.

If you have other questions, please email us at wji3@ualberta.ca.