Page content last modified on 2026-07-01.

UVG-VCM: Benchmarking Dataset for Machine-Oriented Visual Data Compression

UVG-VCM is a new video dataset specifically designed for machine-oriented codec evaluation. The dataset comprises 20 uncompressed and annotated video sequences distributed under the open CC-BY 4.0 license, most of which in 4K, 60 fps, 16 bits YUV444 format. The sequences represent a diverse collection of realistic machine vision use cases, such as object detection, tracking, segmentation, human pose estimation, depth estimation, and license plate recognition. This dataset is intended to foster reproducible benchmarking of future machine oriented video codecs.

Please cite the following paper for any usage of the dataset:

T. Partanen, M. Anttila, R. Kortelahti, G. Gautier, A. Mercat, and J. Vanne, “UVG-VCM: Benchmarking dataset for machine-oriented visual data compression,” Accepted to Int. Conf. Qual. Multimedia Exper., Cardiff, United Kingdom, Jun.–Jul. 2026.

About the annotations

Each sequence contains a YUV video file and machine task annotations. Depth map annotations and instance segmentation masks are provided as per-frame PNG images, while all other annotations are stored in JSON format, including polygon representations of the instance segmentation masks. Visualization is provided for one task for each sequence.

COCO Categories — Object Detection / Tracking / Segmentation (80 classes)
1person
2bicycle
3car
4motorcycle
5airplane
6bus
7train
8truck
9boat
10traffic light
11fire hydrant
12stop sign
13parking meter
14bench
15bird
16cat
17dog
18horse
19sheep
20cow
21elephant
22bear
23zebra
24giraffe
25backpack
26umbrella
27handbag
28tie
29suitcase
30frisbee
31skis
32snowboard
33sports ball
34kite
35baseball bat
36baseball glove
37skateboard
38surfboard
39tennis racket
40bottle
41wine glass
42cup
43fork
44knife
45spoon
46bowl
47banana
48apple
49sandwich
50orange
51broccoli
52carrot
53hot dog
54pizza
55donut
56cake
57chair
58couch
59potted plant
60bed
61dining table
62toilet
63tv
64laptop
65mouse
66remote
67keyboard
68cell phone
69microwave
70oven
71toaster
72sink
73refrigerator
74book
75clock
76vase
77scissors
78teddy bear
79hair drier
80toothbrush
COCO Pose Keypoints (17 points)
  1. nose
  2. left_eye
  3. right_eye
  4. left_ear
  5. right_ear
  6. left_shoulder
  7. right_shoulder
  8. left_elbow
  9. right_elbow
  10. left_wrist
  11. right_wrist
  12. left_hip
  13. right_hip
  14. left_knee
  15. right_knee
  16. left_ankle
  17. right_ankle
DOTA Categories — Oriented Object Detection (15 classes)
0plane
1ship
2storage tank
3baseball diamond
4tennis court
5basketball court
6ground track field
7harbor
8bridge
9large vehicle
10small vehicle
11helicopter
12roundabout
13soccer ball field
14swimming pool

Highway View

Object Detection Tracking Instance Segmentation Panoptic Segmentation

3840x2160 · 60 fps · 600 frames · YUV 444 · 16-bit

Downloadsexpand_more

Floorball Train

Object Detection Tracking Instance Segmentation

3840x2160 · 60 fps · 600 frames · YUV 444 · 16-bit

Downloadsexpand_more

Floorball Game

Object Detection Tracking Instance Segmentation Pose Estimation

3840x2160 · 60 fps · 420 frames · YUV 444 · 16-bit

Downloadsexpand_more

Spot Robot

Object Detection Tracking

3840x2160 · 60 fps · 600 frames · YUV 444 · 16-bit

Downloadsexpand_more

Volleyball Game

Object Detection Tracking Instance Segmentation

3840x2160 · 60 fps · 600 frames · YUV 444 · 16-bit

Downloadsexpand_more

Winter Drive

Object Detection Tracking Instance Segmentation

3840x2160 · 60 fps · 600 frames · YUV 444 · 16-bit

Downloadsexpand_more

Highway Drive

Visualization coming soon
Object Detection Tracking Instance Segmentation Panoptic Segmentation

3840x2160 · 60 fps · 600 frames · YUV 444 · 16-bit

Downloadsexpand_more

Traffic Lights

Object Detection Tracking Instance Segmentation

3840x2160 · 60 fps · 600 frames · YUV 444 · 16-bit

Downloadsexpand_more

Office Walk

Object Detection Tracking Instance Segmentation

3840x2160 · 60 fps · 600 frames · YUV 444 · 16-bit

Downloadsexpand_more

Boat Reverse

Object Detection Tracking Instance Segmentation

3840x2160 · 60 fps · 300 frames · YUV 444 · 16-bit

Downloadsexpand_more

Campus View

Object Detection Tracking

3840x2160 · 60 fps · 600 frames · YUV 444 · 16-bit

Downloadsexpand_more

Job Fair

Object Detection Tracking

3840x2160 · 60 fps · 600 frames · YUV 444 · 16-bit

Downloadsexpand_more

Car Park

License Plate Recognition

3840x2160 · 60 fps · 600 frames · YUV 444 · 16-bit

Downloadsexpand_more

Parking Garage

License Plate Recognition

3840x2160 · 60 fps · 1,140 frames · YUV 444 · 16-bit

Downloadsexpand_more

Karting Time

License Plate Recognition

3840x2160 · 60 fps · 70 frames · YUV 444 · 16-bit

Downloadsexpand_more

Under Park (Stereo)

Depth Estimation

1920x1080 · 30 fps · 300 frames · YUV 422 · 8-bit

Downloadsexpand_more

Auditorium Walk (Stereo)

Depth Estimation

1920x1080 · 30 fps · 300 frames · YUV 422 · 8-bit

Downloadsexpand_more

Synthetic Aerial

Oriented Object Detection

3840x2160 · 60 fps · 300 frames · YUV 444 · 8-bit

Downloadsexpand_more

Synthetic Drone

Object Detection Tracking Instance Segmentation

3840x2160 · 60 fps · 600 frames · YUV 444 · 8-bit

Downloadsexpand_more

Synthetic City

Object Detection Tracking Instance Segmentation

3840x2160 · 60 fps · 600 frames · YUV 444 · 8-bit

Downloadsexpand_more

Disclaimer

All the information and any part thereof provided on this website are provided « AS IS » without warranty of any kind either expressed or implied including, without limitation, warranties of merchantability, fitness for a particular purpose or non infringement of intellectual property rights.

Tampere University makes no representations or warranties as to the accuracy or completeness of any materials and information incorporated thereto and contained on this website. Tampere University makes no representations or warranties that access to this website will be uninterrupted or error-free, that this website (the materials and/or any information incorporated thereto) will be secure and free of virus or other harmful components.