Load the MARS dataset in Python fast. Large video-based individual reidentification dataset. Stream MARS while training ML models in PyTorch and TensorFlow.
Visualization of the MARS dataset on the Activeloop Platform
MARS dataset
What is MARS Dataset?
The MARS (Motion Analysis and Re-distinguishing proof Set) is a large video-based individual reidentification dataset. MARS is an extension of the Market-1501 dataset. The dataset was created by placing six near-synchronized cameras (one of which was a 640*480 SD camera, and the other five cameras were 1,080*1920 HD cameras) on the campus of Tsinghua University. The MARS dataset is made up of 1,261 unique pedestrians that were captured by two or more cameras.
Download MARS Dataset in Python
Instead of downloading the MARS dataset in Python, you can effortlessly load it in Python via our open-source package Hub with just one line of code.
Load MARS Dataset Training Subset in Python
1
import hub
2
ds = hub.load("hub://activeloop/mars-train")
Copied!
Load MARS Dataset Test Subset in Python
1
import hub
2
ds = hub.load("hub://activeloop/mars-test")
Copied!
MARS Dataset Structure
MARS Data Fields
images: tensor to represent image in jpg format.
video_nos: tensor to identify the video to which frame belong to.
frame_nos: tensor to identify each frame within video.
camera_nos: tensor to identify the camera id.
pedestrian_ids: tensor to identify the person with unique id.
Paper: Introduced by Zheng, Liang and Bie, Zhi and Sun, Yifan and Wang, Jingdong and Su, Chi and Wang, Shengjin and Tian, Qi in MARS: A Video Benchmark for Large-Scale Person Re-identification
Point of Contact: N/A
MARS Dataset Curators
X Zheng, Liang and Bie, Zhi and Sun, Yifan and Wang, Jingdong and Su, Chi and Wang, Shengjin and Tian, Qi
MARS Dataset Licensing Information
Hub users may have access to a variety of publicly available datasets. We do not host or distribute these datasets, vouch for their quality or fairness, or claim that you have a license to use the datasets. It is your responsibility to determine whether you have permission to use the datasets under their license.
If you're a dataset owner and do not want your dataset to be included in this library, please get in touch through a GitHub issue. Thank you for your contribution to the ML community!
MARS Dataset Citation Information
1
@proceedings{zheng2016mars,
2
title={MARS: A Video Benchmark for Large-Scale Person Re-identification},
3
author={Zheng, Liang and Bie, Zhi and Sun, Yifan and Wang, Jingdong and Su, Chi and Wang, Shengjin and Tian, Qi},
4
booktitle={European Conference on Computer Vision},
5
year={2016},
6
organization={Springer}
7
}
Copied!
MARS Dataset FAQs
What is the MARS dataset for Python?
MARS (Motion Analysis and Re-identification Set) is a large-scale video-based person reidentification dataset that is a follow-up to the Market-1501 dataset. In addition, the dataset includes 3,248 distractors to make it more realistic. The tracklets were generated automatically using the Deformable Part Model and the GMMCP tracker (mostly 25-50 frames long).
How to download the MARS dataset in Python?
You can load MARS dataset fast with one line of code using the open-source package Activeloop Hub in Python. See detailed instructions on how to load MARS dataset training subset and testing subset in Python.
How can I use MARS dataset in PyTorch or TensorFlow?