A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2020; you can also visit the original URL.
The file type is application/pdf.
Filters
DASOT: A Unified Framework Integrating Data Association and Single Object Tracking for Online Multi-Object Tracking
2020
PROCEEDINGS OF THE THIRTIETH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE AND THE TWENTY-EIGHTH INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE CONFERENCE
Preserved Fulltext
In this paper, we propose an online multi-object tracking (MOT) approach that integrates data association and single object tracking (SOT) with a unified convolutional network (ConvNet), named DASOTNet ...
Data association is treated as a special re-identification task and solved by learning discriminative features for different targets in the data association branch. ...
In this paper, we propose an online MOT algorithm with a neat network that integrates data association and SOT into a unified framework. ...
doi:10.1609/aaai.v34i07.6694
fatcat:vsxn4dqbsnhsjbwz6cvfmgusfy
One More Check: Making "Fake Background" Be Tracked Again
[article]
2021
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2021; you can also visit the original URL.
The file type is application/pdf.
Other Versions
The one-shot multi-object tracking, which integrates object detection and ID embedding extraction into a unified network, has achieved groundbreaking results in recent years. ...
The re-check network innovatively expands the role of ID embedding from data association to motion forecasting by effectively propagating previous tracklets to the current frame with a small overhead. ...
Separated Framework.UMA and DASOT only integrate temporal cues mining and data association into a model, which is separated with the object detector.This is obviously tedious and time-consuming since the ...
arXiv:2104.09441v2
fatcat:lgnfq6p5bfbxthipa2uvnsm5zy
Recent Advances in Embedding Methods for Multi-Object Tracking: A Survey
[article]
2024
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2024; you can also visit the original URL.
The file type is application/pdf.
Other Versions
Multi-object tracking (MOT) aims to associate target objects across video frames in order to obtain entire moving trajectories. ...
In this survey, we first conduct a comprehensive overview with in-depth analysis for embedding methods in MOT from seven different perspectives, including patch-level embedding, single-frame embedding, ...
In addition, DASOT integrates data association and SOT in a unified framework, in which dense correlation feature maps are estimated for the temporal association, built upon truncated ResNet-50 with the ...
arXiv:2205.10766v2
fatcat:hxawibw42nbl3jcb34zncfusey