Pith. sign in

REVIEW 3 major objections 6 minor 70 references

TPT-Bench: A Large-Scale, Long-Term and Robot-Egocentric Dataset for Benchmarking Target Person Tracking

T0 review · 3 major / 6 minor · reviewed 2026-08-15 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read TPT-Bench, a 48-sequence robot-egocentric dataset, shows that current target-person trackers reach only 66.06 F-score and 16.26 zero-false-positive recall in crowded environments.

desk verdict TPT-Bench fills a real gap as a public robot-egocentric long-term tracking dataset, but the annotation-validation evidence is too thin to fully trust the difficulty claims. read the letter →

arxiv 2505.07446 v2 pith:U3ZLKRRC submitted 2025-05-12 cs.RO

Hanjing Ye , Yu Zhan , Weixi Situ , Guangcheng Chen , Jingwen Yu , Ziqi Zhao , Kuanqi Cai , Arash Ajoudani
show 1 more author
Hong Zhang
This is my paper · ORCID
classification cs.RO
keywords targetpersontrackingrobotegocentricvisionlong-termbenchmarkre-identificationfollowingcrowdedenvironmentsmulti-modalsensordatasethuman-robotinteraction
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

TPT-Bench is a new benchmark for tracking a single target person from a mobile robot's egocentric viewpoint in crowded, unstructured environments. The paper's central claim is that long-term target person tracking is far harder than existing benchmarks imply: on its 48 sequences with 571,982 annotated frames, the best evaluated method, RPF-ReID with KPR re-identification features, reaches an F-score of 66.06 and an average max recall at 100% precision of 16.26, compared with up to 99.1 F-score on an older person-following dataset and 70.3 on LaSOT. If the benchmark is sound, it shows that re-identifying a lost target among many pedestrians, rather than short-term occlusion handling, is the unsolved core of robot person following. That matters for assistive robots and embodied agents that must stay with the right person over minutes-long episodes.

What carries the argument

The load-bearing object is the dataset's annotation layer: 571,982 frames of panoramic RGB images with tight 2D bounding boxes, absent-indicator labels, and per-sequence language descriptions covering appearance, lighting, scenario, and clothing changes. These labels convert long-term tracking into a measurable re-identification problem, since disappearance intervals are long enough and crowded enough to make target search necessary. The supporting evaluation machinery is the long-term tracking metric pair from Lukežič et al. 2020: F-score, which balances precision and recall over a confidence threshold, and AMR, average max recall at 100% precision, which penalizes any false positive that would make a robot follow the wrong person. The annotation pipeline also supplies a prior: MOT+Target-ReID trackers are run first, the best visual result is selected as a seed, and annotators refine boxes and absent labels, with 15 Hz annotations interpolated to 30 Hz.

What would settle it

Re-annotate a random sample of sequences from scratch without showing annotators any tracker output, then compare the new boxes to the released ones; if mean intersection-over-union falls below about 0.9 on visible frames, or if re-running the top methods on the re-annotated subset changes their F-score ranking, then the benchmark's difficulty claims rest on annotation bias rather than on the tracking task itself.

Watch

Extended reading notes

Core claim

The paper introduces TPT-Bench as the first large-scale, multi-modal, robot-egocentric target-person-tracking benchmark, and demonstrates that state-of-the-art trackers fail on it. The dataset consists of 5.3 hours of recordings across 48 sequences in crowded indoor and outdoor settings, captured by a human pushing a sensor-equipped cart; annotations are frame-level 2D bounding boxes with explicit absent labels, totaling 571,982 frames. Target disappearances are frequent, averaging 45.4 per sequence and 65.3 seconds absent per sequence, and while the target is gone the field of view contains an average of 9.6 distractor pedestrians. Under long-term tracking metrics, the strongest method, RPF-ReID with KPR features, reaches only 66.06 F-score and 16.26 AMR, while the same method class scores 70.3 F-score on LaSOT and 99.1 F-score on Chen's dataset. The authors conclude that no current method reliably re-identifies a target person after long disappearance in crowded public spaces.

Load-bearing premise

The benchmark's ground-truth boxes are assumed to be accurate and neutral, but they are seeded by the same family of MOT-plus-target-re-identification trackers the paper evaluates, with no published inter-annotator agreement or manual-validation statistics; if those priors carry tracker-specific bias, the measured difficulty and method rankings could shift.

Editorial extensions

If this is right

  • Long-term re-identification is the main bottleneck: targets disappear on average 45.4 times per sequence, for 65.3 seconds in aggregate, and even the best tracker loses them.
  • MOT-plus-target-ReID methods with human priors beat generic single-object trackers by a wide margin, with the best SOT method at 40.05 F-score versus 66.06 for the best overall method.
  • The performance drop relative to older datasets, up to 33 points of F-score versus Chen's dataset, implies that laboratory-style person-following evaluations overstate the readiness of current methods for crowded public spaces.
  • Zero-false-positive recall is especially low for every method, so a robot using these trackers could confidently follow the wrong person; the safest current method still scores only 16.26 AMR.
  • The released multi-modal streams, including panoramic, RGB-D, LiDAR, IMU, and odometry data, provide the substrate for moving beyond vision-only tracking, although only panoramic images are annotated in this version.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The large jump from ResNet18 to KPR features suggests that part-aware, occlusion-robust re-identification, rather than better short-term tracking, is the highest-leverage direction; a direct test would be training a part-based ReID model on these sequences and watching whether the 66 F-score ceiling moves.
  • Because online continual learning adds roughly 12 F-score points to the same method, TPT-Bench is well suited as a stress test for memory-consolidation and replay algorithms under appearance drift, an extension the paper motivates but does not run.
  • The 2.5D ground-plane tool suggests a concrete fusion baseline: using projected LiDAR depth to gate re-identification candidates could reduce the false positives that AMR punishes, since AMR of the best method is only 16.26.
  • The dataset includes person-search episodes, so it can be repurposed beyond tracking as an active-search benchmark, measuring how long a robot takes to locate and re-acquire a disappeared target.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 6 minor

Summary. The paper introduces TPT-Bench, a large-scale, long-term, robot-egocentric dataset for target person tracking (TPT), collected by a human pushing a sensor-equipped cart in crowded indoor and outdoor environments. The dataset includes 48 sequences totaling 5.3 hours, with panoramic, RGB-D, LiDAR, IMU, and odometry data, and 571,982 frames of 2D bounding-box annotations at 30 Hz (interpolated from 15 Hz manual annotations). The authors describe the annotation pipeline, which uses MOT+Target-ReID method outputs as priors and human refinement, and then evaluate a wide range of SOT, MOT+Target-ReID, and Detection+ReID baselines using measures such as F-score and average max recall at 100% precision (AMR). The results show that existing methods perform poorly, with the best method (RPF-ReID with KPR features) reaching an F-score of 66.06 and AMR of 16.26, and the paper claims TPT-Bench is more challenging than LaSOT and Chen's dataset for long-term tracking.

Significance. If the ground-truth annotations are reliable and the evaluation is unbiased, TPT-Bench is a valuable contribution to the HRI, robotics, and computer-vision communities. Its strengths are the large multi-modal dataset, the inclusion of long-term challenges such as target disappearance and re-identification among many distractors, the comprehensive evaluation of current state-of-the-art methods, and the public release of data, code, and development tools. The benchmark fills a clear gap in TPT evaluation, which has been limited to controlled or short-term settings.

major comments (3)
  1. [Sec. 3.2, Annotation and Labels] The ground-truth labels are generated with a prior from the same family of algorithms that are later evaluated: the text states, "we apply all MOT + Target-ReID methods to a given sequence and visually inspect the resulting tracking videos to select the best tracking result as a prior annotation." This introduces a risk of circularity. If the prior method systematically loses the target or mislabels absent frames, the human annotators may unconsciously inherit those errors, and the evaluated MOT+Target-ReID methods (including the authors' own RPF-ReID, Ye et al. 2024) would be favored. The paper provides no quantitative evidence that the final ground truth is independent of the prior, e.g., inter-annotator agreement, comparison between the prior and final labels, or analysis of annotation error rates. This is load-bearing because the central claims that the benchmark is difficult and that MOT+Target-ReID methods outperform SOT methods depend directly on the validity and unbiasedness of the ground truth.
  2. [Sec. 3.2, annotation upsampling] The paper states that annotations are made at 15 Hz and then "upsample the annotations from 15 Hz to 30 Hz using linear interpolation." No validation of this interpolation is reported. Since the evaluation metrics in Sec. 4.1, especially AMR in Eq. (3), rely on IoU thresholds, any distortion of interpolated boxes during fast motion could significantly change the measured performance, potentially inflating the apparent difficulty of the benchmark. The authors should provide an interpolation error analysis (e.g., comparing interpolated boxes against independently annotated 30 Hz boxes on a random subset) and report sensitivity results by running the evaluation on the 15 Hz hand-annotated frames alone.
  3. [Sec. 4.1, evaluation protocol] The definition of AMR in Eq. (3) sums over a set of IoU thresholds, but the exact values of this set are not specified anywhere in the paper. Similarly, the method for selecting the primary F-score ("the highest F-score at an optimal threshold") needs a precise description (grid, step size, per-sequence vs. pooled). Without these details, the reported numbers cannot be independently reproduced, and the cross-dataset comparisons in Table 3 lose their evidential value.
minor comments (6)
  1. [Sec. 3.2, Annotation and Labels] The phrase "tightest up-right one" should be "tightest upright one"; this is a typographical error.
  2. [Table 1, caption] The symbols "%" and "!" in the "Robot Egocentric" column are not explained in the caption or the text; the legend should be made explicit.
  3. [Sec. 4.1, evaluation protocol] The set of IoU thresholds used to compute AMR is unspecified; the authors should list the exact thresholds, e.g., {0.5, 0.75, 0.9} or a range of values.
  4. [Sec. 4.1, evaluation protocol] The statement "using the initial template" is ambiguous: it should clarify whether the trackers are initialized with only the first-frame bounding box or also with the textual sequence descriptions provided in the dataset.
  5. [Sec. 4.5, comparison with other datasets] The comparisons in Table 3 involve datasets with different sequence counts (48 vs. 20 vs. 11); reporting sequence-level standard deviations or confidence intervals would make the claim of higher difficulty more robust.
  6. [Sec. 3.3, dataset statistics] The units "114.3 pixel 2" and "167.4 pixel2" should be rendered as pixels² (or px²) for clarity.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the benchmark's central claims are empirical measurements, and the model-assisted annotation pipeline is a validity concern, not a derivation that reduces to its own inputs.

full rationale

TPT-Bench is an empirical benchmark paper rather than a derivation from first principles. Its central claims—best tracker F-score of 66.06, long disappearance intervals, and the difficulty comparison against LaSOT and Chen's dataset—are measurements made against released annotations and independent external benchmarks. No parameter is fitted to the headline result, no metric is defined in terms of the quantity it is used to predict, and no uniqueness theorem or ansatz is imported from the authors' prior work as the load-bearing justification. The annotation pipeline in Sec. 3.2 does bootstrap a prior from MOT+Target-ReID outputs, and the absence of inter-annotator agreement or manual-validation statistics is a legitimate quality weakness that could affect ground-truth validity and tracker rankings. However, the paper's own text describes human refinement of every frame, explicit correction of false-absent labels by selecting the target ID from tracked boxes, and manual drawing of missing boxes; the final labels are therefore not forced to equal any single tracker's output by construction. The main conclusions also include evaluations on LaSOT and Chen's dataset, which are independent of the TPT-Bench annotation bootstrap. Self-citations such as Ye et al. (2024) appear as evaluated baselines, not as unverified justifications of the central claims. Method-assisted annotation is common benchmark practice and, absent a demonstrated equation-level reduction of the benchmark result to the annotation prior, this is a validity risk rather than circularity. Accordingly, no circular derivation step is present and the score is 0.

Assumptions & free parameters 0 free parameters · 3 assumptions · 0 invented entities

No free parameters are fit in this dataset paper; the central claims rest on the validity of the annotations and the assumption that the push-cart view approximates robot egocentric observation.

assumptions (3)
  • domain assumption Push-cart egocentric observations are representative of mobile robot egocentric perception.
    The platform is a human-pushed cart with sensors mounted at robot-like heights (Sec 3.1.1); the paper argues this captures human-like following and robot kinematic constraints, but no robot deployment demonstrates equivalence.
  • domain assumption The semi-automatically generated and interpolated 2D bounding boxes are accurate ground truth.
    Sec 3.2 uses tracker outputs as prior annotations, human refinement, and linear interpolation from 15 Hz to 30 Hz; no annotation error rate or inter-annotator agreement is reported.
  • domain assumption Selection of the best tracking result as annotation prior does not introduce bias toward the evaluated methods.
    Sec 3.2: 'we apply all MOT + Target-ReID methods to a given sequence and visually inspect the resulting tracking videos to select the best tracking result as a prior annotation.' If this prior comes from one of the benchmarked methods (e.g., RPF-ReID), evaluation rankings can be inflated for that family.

how reviews work

0 comments
Cite this review

Pith. "Pith review of TPT-Bench: A Large-Scale, Long-Term and Robot-Egocentric Dataset for Benchmarking Target Person Tracking." pith.science (2026). https://pith.science/paper/U3ZLKRRC

@misc{pith2026250507446,
  author       = {Pith},
  title        = {Pith review of: TPT-Bench: A Large-Scale, Long-Term and Robot-Egocentric Dataset for Benchmarking Target Person Tracking},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/U3ZLKRRC}},
  note         = {Machine review of arXiv:2505.07446}
}
read the original abstract

Tracking a target person from robot-egocentric views is crucial for developing autonomous robots that provide continuous personalized assistance or collaboration in Human-Robot Interaction (HRI) and Embodied AI. However, most existing target person tracking (TPT) benchmarks are limited to controlled laboratory environments with few distractions, clean backgrounds, and short-term occlusions. In this paper, we introduce a large-scale dataset designed for TPT in crowded and unstructured environments, demonstrated through a robot-person following task. The dataset is collected by a human pushing a sensor-equipped cart while following a target person, capturing human-like following behavior and emphasizing long-term tracking challenges, including frequent occlusions and the need for re-identification from numerous pedestrians. It includes multi-modal data streams, including odometry, 3D LiDAR, IMU, panoramic images, and RGB-D images, along with exhaustively annotated 2D bounding boxes of the target person across 48 sequences, both indoors and outdoors. Using this dataset and visual annotations, we perform extensive experiments with existing SOTA TPT methods, offering a thorough analysis of their limitations and suggesting future research directions.

Figures

Figures reproduced from arXiv: 2505.07446 by the authors.

Figure 1
Figure 1. Example robot-egocentric visual sequences and annotations of the TPT-Bench dataset. [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. The platform and sensor suite. A panoramic camera, depth camera, and 3-D LiDAR are fixed on a push-cart in positions that mirror the typical sensor layout of most mobile robots [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Third-view examples of dataset collection [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (8 more)
Figure 4
Figure 4. Figure 4: Four example sequences from the TPT-Bench dataset. [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: The description example. The annotated description involves the basic attributes, the target person’s appearance at the initial frame, the scenarios involved, the lighting conditions, and whether a clothing change occurs [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 6
Figure 6. Figure 6: Examples of different lighting conditions. [PITH_FULL_IMAGE:figures/full_fig_p006_6.png]
Figure 7
Figure 7. Figure 7: The distribution of scenario category counts. [PITH_FULL_IMAGE:figures/full_fig_p007_7.png]
Figure 9
Figure 9. Figure 9: The histogram of dataset statistics. In the “disappeared intervals” subfigure, durations over 30 seconds are excluded since they are considered part of the search procedure. The blue line represents the median [PITH_FULL_IMAGE:figures/full_fig_p007_9.png]
Figure 10
Figure 10. Figure 10: The dataset organization. a mean of 9.6 distractors and sometimes more than 20, underscoring the difficulty of re-identification in crowded scenes. Finally, the median target bounding-box area is roughly 114.3 pixel2 , with a per-sequence variance of around 167.4 pixe…
Figure 11
Figure 11. Figure 11: Development tool usages. (a) The projected point cloud overlaid onto panoramic images using our custom development tools, with point colors encoding relative distances. This process involves calibration, annotation loaders, and point cloud processing. (b) The 2.5D gro…
Figure 12
Figure 12. Figure 12: Failure cases. The upper-left number shows the estimated confidence of the target and the upper-right number shows the frame number. The green triangle and box represent the ground truth, while the blue box shows the estimation. For the third subfigure of (b), the tar…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

70 extracted references · 58 canonical work pages

  1. [1]

    In: Proc

    Bhat G, Danelljan M, Gool LV and Timofte R (2019) Learning discriminative model prediction for tracking. In: Proc. IEEE Int. Conf. Comput. Vis. pp. 6182--6191

    work page 2019

  2. [2]

    IEEE Trans

    C ehovin L, Leonardis A and Kristan M (2016) Visual object tracking performance measures revisited. IEEE Trans. Image Process. 25(3): 1261--1274

    work page 2016

  3. [3]

    Chaudhry A, Rohrbach M, Elhoseiny M, Ajanthan T, Dokania PK, Torr PH and Ranzato M (2019) Continual learning with tiny episodic memories. In: Int. Conf. Mach. Learn. PMLR

    work page 2019

  4. [4]

    Chen BX, Sahdev R and Tsotsos JK (2017 a ) Integrating stereo vision with a cnn tracker for a person-following robot. In: Int. Conf. Comput. Vis. Syst. (ICVS). Springer, pp. 300--313

    work page 2017

  5. [5]

    In: Conf

    Chen BX, Sahdev R and Tsotsos JK (2017 b ) Person following robot using selected online ada-boosting with stereo camera. In: Conf. Comput. Rob. Vis. (CRV). pp. 48--55. doi:10.1109/CRV.2017.55

    work page doi:10.1109/crv.2017.55 2017

  6. [6]

    In: Proc

    Cui Y, Jiang C, Wang L and Wu G (2022) Mixformer: End-to-end tracking with iterative mixed attention. In: Proc. IEEE Conf. Comput. Vis. Pattern Recognit. pp. 13608--13618

    work page 2022

  7. [7]

    IEEE Trans

    Cui Y, Jiang C, Wu G and Wang L (2024) Mixformer: End-to-end tracking with iterative mixed attention. IEEE Trans. Pattern Anal. Mach. Intell

    work page 2024

  8. [8]

    In: Proc

    Dai K, Zhang Y, Wang D, Li J, Lu H and Yang X (2020) High-performance long-term tracking with meta-updater. In: Proc. IEEE Conf. Comput. Vis. Pattern Recognit. pp. 6298--6307

    work page 2020

Show all 70 references
  1. [9]

    arXiv preprint arXiv:2003.09003

    Dendorfer P (2020) Mot20: A benchmark for multi object tracking in crowded scenes. arXiv preprint arXiv:2003.09003

    2020 arXiv

  2. [10]

    IEEE Access

    Eirale A, Martini M and Chiaberge M (2025) Human following and guidance by autonomous mobile robots: A comprehensive review. IEEE Access

    2025

  3. [11]

    In: Proc

    Eisenbach M, L \"u bberstedt J, Aganian D and Gross HM (2023) A little bit attention is all you need for person re-identification. In: Proc. IEEE Int. Conf. Robot. Autom. IEEE, pp. 7598--7605

    2023

  4. [12]

    (2021) Lasot: A high-quality large-scale single object tracking benchmark

    Fan H, Bai H, Lin L, Yang F, Chu P, Deng G, Yu S, Harshit, Huang M, Liu J et al. (2021) Lasot: A high-quality large-scale single object tracking benchmark. Int. J. Comput. Vis. 129: 439--461

    2021

  5. [13]

    (2025) Principles and guidelines for evaluating social robot navigation algorithms

    Francis A, P \'e rez-d’Arpino C, Li C, Xia F, Alahi A, Alami R, Bera A, Biswas A, Biswas J, Chandra R et al. (2025) Principles and guidelines for evaluating social robot navigation algorithms. ACM Transactions on Human-Robot Interaction 14(2): 1--65

    2025

  6. [14]

    arXiv preprint arXiv:2107.08430

    Ge Z (2021) Yolox: Exceeding yolo series in 2021. arXiv preprint arXiv:2107.08430

    2021 arXiv

  7. [15]

    In: IEEE Int

    Goldhoorn A, Garrell A, Alqu \'e zar R and Sanfeliu A (2014) Continuous real time pomcp to find-and-follow people by a humanoid service robot. In: IEEE Int. Conf. Humanoid Robots . IEEE, pp. 741--747

    2014

  8. [16]

    (2017) Roreas: robot coach for walking and orientation training in clinical post-stroke rehabilitation—prototype implementation and evaluation in field trials

    Gross HM, Scheidig A, Debes K, Einhorn E, Eisenbach M, Mueller S, Schmiedel T, Trinh TQ, Weinrich C, Wengefeld T et al. (2017) Roreas: robot coach for walking and orientation training in clinical post-stroke rehabilitation—prototype implementation and evaluation in field trial...

    2017

  9. [17]

    IEEE Trans

    Honig SS, Oron-Gilad T, Zaichyk H, Sarne-Fleischmann V, Olatunji S and Edan Y (2018) Toward socially aware person-following robots. IEEE Trans. Cogn. Dev. Syst. 10(4): 936--954

    2018

  10. [18]

    In: AAAI Conf

    Huang L, Zhao X and Huang K (2020) Globaltrack: A simple and strong baseline for long-term tracking. In: AAAI Conf. Artif. Intell., volume 34. pp. 11037--11044

    2020

  11. [19]

    Islam MJ, Hong J and Sattar J (2019) Person-following by autonomous robots: A categorical overview. Int. J. Robot. Res. (IJRR) 38(14): 1581--1618

    2019

  12. [20]

    IEEE Trans

    Javed S, Danelljan M, Khan FS, Khan MH, Felsberg M and Matas J (2022) Visual object tracking with discriminative filters and siamese networks: a survey and outlook. IEEE Trans. Pattern Anal. Mach. Intell. 45(5): 6552--6574

    2022

  13. [21]

    IEEE Robot

    Karnan H, Nair A, Xiao X, Warnell G, Pirk S, Toshev A, Hart J, Biswas J and Stone P (2022) Socially compliant navigation dataset (scand): A large-scale dataset of demonstrations for social navigation. IEEE Robot. Autom. Lett. 7(4): 11807--11814

    2022

  14. [22]

    Karunarathne D, Morales Y, Kanda T and Ishiguro H (2018) Model of side-by-side walking without the robot knowing the goal. Int. J. Social Robot. (IJSR) 10: 401--420

    2018

  15. [23]

    Koide K, Miura J and Menegatti E (2020) Monocular person tracking and identification with on-line deep feature selection for person following robots. Robot. and Auton. Syst. 124: 103348

    2020

  16. [24]

    In: Proc

    Leigh A, Pineau J, Olmedo N and Zhang H (2015) Person tracking and following with 2d laser scanners. In: Proc. IEEE Int. Conf. Robot. Autom. pp. 726--733. doi:10.1109/ICRA.2015.7139259

    2015

  17. [25]

    In: Proc

    Li B, Wu W, Wang Q, Zhang F, Xing J and Yan J (2019) Siamrpn++: Evolution of siamese visual tracking with very deep networks. In: Proc. IEEE Conf. Comput. Vis. Pattern Recognit. pp. 4282--4291

    2019

  18. [26]

    (2024) Deepseek-v3 technical report

    Liu A, Feng B, Xue B, Wang B, Wu B, Lu C, Zhao C, Deng C, Zhang C, Ruan C et al. (2024) Deepseek-v3 technical report. arXiv preprint arXiv:2412.19437

    2024 arXiv

  19. [27]

    IEEE Trans

    Luke \'z i c A, Zajc L C , Voj \' r T, Matas J and Kristan M (2020) Performance evaluation methodology for long-term single-object tracking. IEEE Trans. Cybern. 51(12): 6305--6318

    2020

  20. [28]

    Artificial intelligence 293: 103448

    Luo W, Xing J, Milan A, Zhang X, Liu W and Kim TK (2021) Multiple object tracking: A literature review. Artificial intelligence 293: 103448

    2021

  21. [29]

    Neurocomputing 469: 28--51

    Mai Z, Li R, Jeong J, Quispe D, Kim H and Sanner S (2022) Online continual learning in image classification: An empirical survey. Neurocomputing 469: 28--51

    2022

  22. [30]

    In: Proc

    Manen S, Gygli M, Dai D and Van Gool L (2017) Pathtrack: Fast trajectory annotation with path supervision. In: Proc. IEEE Int. Conf. Comput. Vis. pp. 290--299

    2017

  23. [31]

    IEEE Trans

    Martin-Martin R, Patel M, Rezatofighi H, Shenoi A, Gwak J, Frankel E, Sadeghian A and Savarese S (2021) Jrdb: A dataset and benchmark of egocentric robot visual perception of humans in built environments. IEEE Trans. Pattern Anal. Mach. Intell. 45(6): 6748--6765

    2021

  24. [32]

    In: Proc

    Mayer C, Danelljan M, Bhat G, Paul M, Paudel DP, Yu F and Van Gool L (2022) Transforming model prediction for tracking. In: Proc. IEEE Conf. Comput. Vis. Pattern Recognit. pp. 8731--8740

    2022

  25. [33]

    In: Proc

    Mayer C, Danelljan M, Paudel DP and Van Gool L (2021) Learning target candidate association to keep track of what not to track. In: Proc. IEEE Int. Conf. Comput. Vis. pp. 13444--13454

    2021

  26. [34]

    In: Proc

    Mayer C, Danelljan M, Yang MH, Ferrari V, Van Gool L and Kuznetsova A (2024) Beyond sot: Tracking multiple generic objects at once. In: Proc. IEEE Winter Conf. App. Comput. Vis. (WACV) . pp. 6826--6836

    2024

  27. [35]

    In: Proc

    Miao J, Wu Y, Liu P, Ding Y and Yang Y (2019) Pose-guided feature alignment for occluded person re-identification. In: Proc. IEEE Int. Conf. Comput. Vis. pp. 542--551

    2019

  28. [36]

    arXiv preprint arXiv:1603.00831

    Milan A (2016) Mot16: A benchmark for multi-object tracking. arXiv preprint arXiv:1603.00831

    2016 arXiv

  29. [37]

    Morales Y, Miyashita T and Hagita N (2017) Social robotic wheelchair centered on passenger and pedestrian comfort. Robot. and Auton. Syst. 87: 355--362

    2017

  30. [38]

    In: Proc

    Muller M, Bibi A, Giancola S, Alsubaihi S and Ghanem B (2018) Trackingnet: A large-scale dataset and benchmark for object tracking in the wild. In: Proc. Eur. Conf. Comput. Vis. pp. 300--317

    2018

  31. [39]

    In: Proc

    Nguyen DM, Nazeri M, Payandeh A, Datar A and Xiao X (2023) Toward human-like social robot navigation: A large-scale, multi-modal, social human navigation dataset. In: Proc. IEEE/RJS Int. Conf. Intell. Robots Syst. IEEE, pp. 7442--7447

    2023

  32. [40]

    ://openai.com/research/gpt-4o

    OpenAI (2025) Gpt-4o: A large-scale language model. ://openai.com/research/gpt-4o

    2025

  33. [41]

    (2024) Sam 2: Segment anything in images and videos

    Ravi N, Gabeur V, Hu YT, Hu R, Ryali C, Ma T, Khedr H, R \"a dle R, Rolland C, Gustafson L et al. (2024) Sam 2: Segment anything in images and videos. arXiv preprint arXiv:2408.00714

    2024 arXiv

  34. [42]

    In: IROS

    Repiso E, Garrell A and Sanfeliu A (2018) Robot approaching and engaging people in a human-robot companion framework. In: IROS. IEEE, pp. 8200--8205

    2018

  35. [43]

    Repiso E, Garrell A and Sanfeliu A (2020 a ) Adaptive side-by-side social robot navigation to approach and interact with people. Int. J. Social Robot. (IJSR) 12(4): 909--930

    2020

  36. [44]

    RAL 5(2): 2387--2394

    Repiso E, Garrell A and Sanfeliu A (2020 b ) People's adaptive side-by-side model evolved to accompany groups of people by social robots. RAL 5(2): 2387--2394

    2020

  37. [45]

    In: 2023 IEEE International Conference on Advanced Robotics and Its Social Impacts (ARSO)

    Rollo F, Zunino A, Raiola G, Amadio F, Ajoudani A and Tsagarakis N (2023) Followme: A robust person following framework based on visual re-identification and gestures. In: 2023 IEEE International Conference on Advanced Robotics and Its Social Impacts (ARSO). IEEE, pp. 84--89

    2023

  38. [46]

    In: Proc

    Rollo F, Zunino A, Tsagarakis N, Hoffman EM and Ajoudani A (2024) Continuous adaptation in person re-identification for robotic assistance. In: Proc. IEEE Int. Conf. Robot. Autom. pp. 425--431. doi:10.1109/ICRA57147.2024.10611226

    2024

  39. [47]

    Singamaneni PT, Bachiller-Burgos P, Manso LJ, Garrell A, Sanfeliu A, Spalanzani A and Alami R (2024) A survey on socially aware robot navigation: Taxonomy and future challenges. Int. J. Robot. Res. (IJRR) : 02783649241230562

    2024

  40. [48]

    IJRR 41(7): 706--720

    Siva S and Zhang H (2022) Robot perceptual adaptation to environment changes for long-term human teammate following. IJRR 41(7): 706--720. doi:10.1177/0278364919896625

    2022 doi

  41. [49]

    In: Proc

    Somers V, Alahi A and Vleeschouwer CD (2025) Keypoint promptable re-identification. In: Proc. Eur. Conf. Comput. Vis. Springer, pp. 216--233

    2025

  42. [50]

    Tang H, Liang KJ, Grauman K, Feiszli M and Wang W (2024) Egotracks: A long-term egocentric visual object tracking dataset. Adv. Neural Inf. Process. Syst. 36

    2024

  43. [51]

    (2016) Spencer: A socially aware service robot for passenger guidance and help in busy airports

    Triebel R, Arras K, Alami R, Beyer L, Breuers S, Chatila R, Chetouani M, Cremers D, Evers V, Fiore M et al. (2016) Spencer: A socially aware service robot for passenger guidance and help in busy airports. In: Proc. 10th Int. Conf. Field Serv. Robot. (FSR). Springer, pp. 607--622

    2016

  44. [52]

    In: Proc

    Voigtlaender P, Luiten J, Torr PH and Leibe B (2020) Siam r-cnn: Visual tracking by re-detection. In: Proc. IEEE Conf. Comput. Vis. Pattern Recognit. pp. 6578--6588

    2020

  45. [53]

    arXiv preprint arXiv:2505.23189

    Wang S, Zhang J, Li M, Liu J, Li A, Wu K, Zhong F, Yu J, Zhang Z and Wang H (2025) Trackvla: Embodied visual tracking in the wild. arXiv preprint arXiv:2505.23189

    2025 arXiv

  46. [54]

    IJRR : 02783649241303525

    Wei H, Jiao J, Hu X, Yu J, Xie X, Wu J, Zhu Y, Liu Y, Wang L and Liu M (2024) Fusionportablev2: A unified multi-sensor dataset for generalized slam across diverse platforms and scalable environments. IJRR : 02783649241303525

    2024

  47. [55]

    In: Proc

    Wu Y, Lim J and Yang MH (2013) Online object tracking: A benchmark. In: Proc. IEEE Conf. Comput. Vis. Pattern Recognit

    2013

  48. [56]

    In: Proc

    Yan B, Peng H, Fu J, Wang D and Lu H (2021) Learning spatio-temporal transformer for visual tracking. In: Proc. IEEE Int. Conf. Comput. Vis. pp. 10448--10457

    2021

  49. [57]

    In: CVPR

    Yang L, Kang B, Huang Z, Xu X, Feng J and Zhao H (2024) Depth anything: Unleashing the power of large-scale unlabeled data. In: CVPR

    2024

  50. [58]

    In: Proc

    Ye H, Zhao J, Pan Y, Chen W, He L and Zhang H (2023) Robot person following under partial occlusion. In: Proc. IEEE Int. Conf. Robot. Autom. IEEE, pp. 7591--7597

    2023

  51. [59]

    IEEE Robot

    Ye H, Zhao J, Zhan Y, Chen W, He L and Zhang H (2024) Person re-identification for robot person following with online continual learning. IEEE Robot. Autom. Lett

    2024

  52. [60]

    IEEE Trans

    Ye M, Shen J, Lin G, Xiang T, Shao L and Hoi SC (2021) Deep learning for person re-identification: A survey and outlook. IEEE Trans. Pattern Anal. Mach. Intell. 44(6): 2872--2893

    2021

  53. [61]

    In: IROS

    Yoshimi T, Nishiyama M, Sonoura T, Nakamoto H, Tokura S, Sato H, Ozaki F, Matsuhira N and Mizoguchi H (2006) Development of a person following robot with vision based target detection. In: IROS. IEEE, pp. 5286--5291

    2006

  54. [62]

    In: Proc

    Yu J, Ye H, Jiao J, Tan P and Zhang H (2024) Gv-bench: Benchmarking local feature matching for geometric verification of long-term loop closure detection. In: Proc. IEEE/RJS Int. Conf. Intell. Robots Syst. IEEE, pp. 7922--7928

    2024

  55. [63]

    In: Proc

    Yu T, Li D, Yang Y, Hospedales TM and Xiang T (2019) Robust person re-identification by modelling feature uncertainty. In: Proc. IEEE Int. Conf. Comput. Vis. pp. 552--561

    2019

  56. [64]

    IEEE Trans

    Zhang M, Liu X, Xu D, Cao Z and Yu J (2019) Vision-based target-following guider for mobile robot. IEEE Trans. Ind. Electron. 66(12): 9360--9371

    2019

  57. [65]

    Zhang Y, Sun P, Jiang Y, Yu D, Weng F, Yuan Z, Luo P, Liu W and Wang X (2022) Bytetrack: Multi-object tracking by associating every detection box. Proc. Eur. Conf. Comput. Vis

    2022

  58. [66]

    In: 2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)

    Zhao J, Ye H, Zhan Y, Luan H and Zhang H (2024) Human orientation estimation under partial observation. In: 2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). IEEE, pp. 11544--11551

    2024

  59. [67]

    , " * write output.state after.block = add.period write newline

    ENTRY address author booktitle chapter doi edition editor eid howpublished institution isbn journal key month note number organization pages publisher school series title type url volume year label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence...

  60. [68]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

  61. [69]

    , " * write output.state after.block = add.period write newline

    ENTRY address archive author booktitle chapter doi edition editor eid eprint howpublished institution isbn journal key month note number organization pages publisher school series title type url volume year label INTEGERS output.state before.all mid.sentence after.sentence aft...

  62. [70]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.