Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

VastTrack: Vast Category Visual Object Tracking

Liang Peng, Junyuan Gao, Xinran Liu, Weihong Li, Shaohua Dong, Zhipeng Zhang, Heng Fan, Libo Zhang

TL;DR

The results, not surprisingly, show significant drops compared to those on current datasets due to lack of abundant categories and videos from diverse scenarios for training, and more efforts are required to improve general tracking.

Abstract

In this paper, we introduce a novel benchmark, dubbed VastTrack, towards facilitating the development of more general visual tracking via encompassing abundant classes and videos. VastTrack possesses several attractive properties: (1) Vast Object Category. In particular, it covers target objects from 2,115 classes, largely surpassing object categories of existing popular benchmarks (e.g., GOT-10k with 563 classes and LaSOT with 70 categories). With such vast object classes, we expect to learn more general object tracking. (2) Larger scale. Compared with current benchmarks, VastTrack offers 50,610 sequences with 4.2 million frames, which makes it to date the largest benchmark regarding the number of videos, and thus could benefit training even more powerful visual trackers in the deep learning era. (3) Rich Annotation. Besides conventional bounding box annotations, VastTrack also provides linguistic descriptions for the videos. The rich annotations of VastTrack enables development of both the vision-only and the vision-language tracking. To ensure precise annotation, all videos are manually labeled with multiple rounds of careful inspection and refinement. To understand performance of existing trackers and to provide baselines for future comparison, we extensively assess 25 representative trackers. The results, not surprisingly, show significant drops compared to those on current datasets due to lack of abundant categories and videos from diverse scenarios for training, and more efforts are required to improve general tracking. Our VastTrack and all the evaluation results will be made publicly available https://github.com/HengLan/VastTrack.

VastTrack: Vast Category Visual Object Tracking

TL;DR

The results, not surprisingly, show significant drops compared to those on current datasets due to lack of abundant categories and videos from diverse scenarios for training, and more efforts are required to improve general tracking.

Abstract

In this paper, we introduce a novel benchmark, dubbed VastTrack, towards facilitating the development of more general visual tracking via encompassing abundant classes and videos. VastTrack possesses several attractive properties: (1) Vast Object Category. In particular, it covers target objects from 2,115 classes, largely surpassing object categories of existing popular benchmarks (e.g., GOT-10k with 563 classes and LaSOT with 70 categories). With such vast object classes, we expect to learn more general object tracking. (2) Larger scale. Compared with current benchmarks, VastTrack offers 50,610 sequences with 4.2 million frames, which makes it to date the largest benchmark regarding the number of videos, and thus could benefit training even more powerful visual trackers in the deep learning era. (3) Rich Annotation. Besides conventional bounding box annotations, VastTrack also provides linguistic descriptions for the videos. The rich annotations of VastTrack enables development of both the vision-only and the vision-language tracking. To ensure precise annotation, all videos are manually labeled with multiple rounds of careful inspection and refinement. To understand performance of existing trackers and to provide baselines for future comparison, we extensively assess 25 representative trackers. The results, not surprisingly, show significant drops compared to those on current datasets due to lack of abundant categories and videos from diverse scenarios for training, and more efforts are required to improve general tracking. Our VastTrack and all the evaluation results will be made publicly available https://github.com/HengLan/VastTrack.
Paper Structure (18 sections, 14 figures, 5 tables)

This paper contains 18 sections, 14 figures, 5 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Visual Tracking Benchmarks
  4. Visual Tracking Algorithms
  5. Other Vision Benchmarks with Vast Categories
  6. The Proposed VastTrack
  7. Construction Principle
  8. Data Acquisition
  9. Annotation
  10. Attributes
  11. Dataset Split and Evaluation Protocol
  12. Experiments
  13. Evaluation Methodology
  14. Evaluated Trackers
  15. Evaluation Results
  16. ...and 3 more sections

Figures (14)

  • Figure 1: Summary of representative benchmarks, comprising OTB-2013/2015 wu2013onlinewu2015object, TC-128 liang2015encoding, UAV123 mueller2016benchmark, NUS-PRO li2016nus, UAV20L mueller2016benchmark, VOT-2017 kristan2016novel, OxUvA valmadre2018long, GOT-10k huang2019got, TrackingNet muller2018trackingnet, and VastTrack. We can clearly see that VastTrack is larger than all other datasets by containing 2,115 object categories and 50,803 videos. Best viewed in color for all figures in paper.
  • Figure 2: VastTrack, a new large-scale benchmark for facilitating general single object tracking with abundant object categories and videos. Here we show the partial target trajectory in a video. Note, only a very small part of categories and videos are displayed.
  • Figure 3: The number of videos in each object class forms a long-tail distribution, which is common and universal in our real world.
  • Figure 4: Visualization of several annotation examples in the proposed VastTrack.
  • Figure 5: Distribution of videos per attribute.
  • ...and 9 more figures