Pith. sign in

REVIEW 4 major objections 5 minor 44 references

Light-Weight Cross-Modal Enhancement Method with Benchmark Construction for UAV-based Open-Vocabulary Object Detection

T0 review · 4 major / 5 minor · reviewed 2026-08-05 · deepseek-v4-flash

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

Pith's one-line read The paper argues that open-vocabulary detection on drone imagery can be made practical by combining domain-specific UAV pretraining data with a lightweight cross-modal fusion module, and supports this with two new benchmarks and a CAGE modu

desk verdict Useful UAV dataset benchmark, but the +5.3 mAP headline conflates dataset and module gains and the LVLM-generated labels need validation. read the letter →

arxiv 2509.06011 v2 pith:ZUJVQMDL submitted 2025-09-07 cs.CV

Zhenhai Weng , Xinjie Li , Can Wu , Weijie He , Jianfeng Lv , Dong Zhou , Zhongliang Yu This is my paper
classification cs.CV
keywords open-vocabularyobjectdetectionUAVimagerydatasetconstructionvision-languagepretrainingcross-modalfusionzero-shotpseudo-labelingreal-time
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

The paper argues that open-vocabulary object detection fails on drone imagery not because the models are weak but because they are trained on ground-level data. To close that gap, it builds two large UAV-view datasets—UAVDE-2M with 2.4M instances across 1,853 categories and UAVCAP-15K with richly captioned image-text pairs—using an automated label engine, and adds a lightweight cross-attention gated enhancement (CAGE) module to YOLO-World-v2. On VisDrone zero-shot evaluation, the combined contribution raises mAP from 8.59 to 13.9 for the large model while cutting parameters by about 29% and GFLOPs by about 30%, and it transfers to a different remote-sensing benchmark. The authors claim this makes practical, real-time open-vocabulary detection feasible on resource-limited UAV platforms.

What carries the argument

The load-bearing mechanism is the CAGE module: a dual-path fusion block whose cross-attention path grounds textual tokens into spatial image features, a learned per-pixel gate decides where text context is relevant so irrelevant or occluded regions are not polluted, a global FiLM path shifts and scales the whole feature map from a pooled text vector, and a residual connection preserves the original visual stream. Its role is to make a real-time detector align text and vision more cheaply than the T-CSPLayer blocks it replaces, while the datasets supply the aerial-domain priors the module needs.

What would settle it

Re-annotate a random sample of about 1,000 images from the COD and FOP subsets of UAVDE-2M by human experts and measure label agreement with the Qwen2.5-VL annotations. Then train identical YOLO-World-v2 models on the full auto-labeled set and on the human-audited subset: if VisDrone mAP is unchanged while agreement is low, the auto-labeled data are not actually supplying the claimed supervision.

Watch

Extended reading notes

Core claim

The central claim is that the domain gap between ground-level and UAV imagery can be overcome by domain-specific pretraining data plus a cheap fusion module, without enlarging the detector. The paper introduces a refined UAV-Label Engine that turns existing drone and remote-sensing datasets into standardized detection annotations, treating partially-annotated and ambiguously-labeled data as a distinct category and using Qwen2.5-VL for fine-grained re-classification; this yields UAVDE-2M. A second dataset, UAVCAP-15K, supplies long-form captions for vision-language pretraining. The CAGE module replaces the text-vision fusion blocks in YOLO-World-v2's neck with a dual-path design: multi-head c

Load-bearing premise

The whole pipeline depends on the assumption that the labels generated by Qwen2.5-VL for the unlabeled and partially labeled subsets of UAVDE-2M are accurate and consistent enough to train a detector that generalizes to real UAV scenes.

Editorial extensions

If this is right

  • Domain-specific pretraining on UAV-view data is the dominant source of zero-shot gain: adding UAVDE-2M alone lifts VisDrone mAP from 4.58 to 9.22, and UAVCAP-15K adds another 0.38.
  • CAGE improves accuracy at lower cost: at the L scale, mAP rises from 12.2 to 13.9 while parameters drop from 48M to 34M and GFLOPs from 204.5 to 144.0.
  • The approach transfers across domains: models pretrained on UAV data reach 11.2 mAP on SIMD versus 8.93 for general-purpose pretraining.
  • The trained model runs in real time on the tested embedded platform, achieving 22.9 ms per frame compared with 25.38 ms for the baseline.
  • Because CAGE is a drop-in replacement for the original fusion blocks, it can be transplanted to other vision-language detection frameworks without architectural changes.

Reading between the lines

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

  • Beyond the paper: because the ablation attributes most of the gain to the dataset rather than the module, a lighter path to similar results would be to pretrain an existing open-vocabulary detector on UAVDE-2M alone and skip architectural changes when deployment simplicity matters more than peak accuracy.
  • The gate's per-pixel relevance map could be repurposed as a signal for occlusion or domain shift, since it learns where text context is trusted; one testable extension is to check whether its activations correlate with small-object density in aerial scenes.
  • The same UAV-Label Engine pipeline could be pointed at other under-annotated video corpora, such as satellite or underwater footage, to bootstrap open-vocabulary detectors for those domains; this is an extension the paper does not test.
  • A direct test of pseudo-label quality would be to train a model on a human-audited subset of UAVDE-2M and compare its VisDrone mAP to the model trained on the full auto-labeled set; if the numbers match despite low human-model label agreement, the LVLM annotations are not actually carrying the claimed benefit.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The paper proposes a complete UAV-domain open-vocabulary object detection (OVD) solution, consisting of two new benchmarks (UAVDE-2M and UAVCAP-15K) built with a semi-automated 'UAV-Label Engine', and a lightweight Cross-Attention Gated Enhancement (CAGE) module that is inserted into the YOLO-World-v2 neck as a drop-in replacement for T-CSPLayer. The authors report that the combined method improves zero-shot VisDrone mAP from 8.59 to 13.9 for the L-scale model, while reducing parameters and GFLOPs, and shows cross-domain gains on the SIMD remote-sensing benchmark. They also present real-world deployment results on a Jetson Orin NX.

Significance. If the underlying data are reliable and the reported gains are correctly attributed, this is a useful engineering contribution to UAV-focused OVD. The scale of the proposed datasets (2.4M instances, 1,853 categories) and the inclusion of a caption dataset for pretraining are valuable resources for the community. The paper also contains a real-world embedded deployment and latency measurements, which are often missing from OVD papers. However, the central dataset contribution rests on unvalidated LVLM-generated pseudo-labels, and the headline +5.3 mAP conflates dataset and module effects. These issues substantially temper the significance unless the authors can provide additional validation and clearer attribution.

major comments (4)
  1. [Section 3.1 / Table 1] The COD and FOP subsets of UAVDE-2M are annotated using the Qwen2.5-VL LVLM pipeline, but no quantitative validation of these pseudo-labels is provided. There is no precision/recall measurement against source datasets' original annotations, no human-agreement study, and no sample-based error analysis. The class counts for COD datasets (e.g., DenseUAV 748, ERA 838, FloodNet 772) are entirely produced by the LVLM and are never checked. Since these pseudo-labels are a central contribution and are claimed to be 'clean yet exhaustive', the benchmark's reliability is unestablished. The authors should provide a label-quality study, e.g., comparing a random sample against human annotations or the original dataset labels, with per-dataset accuracy/agreement numbers.
  2. [Abstract / Conclusion vs. Table 2] The abstract and conclusion attribute the +5.3 mAP VisDrone improvement to CAGE ('By embedding CAGE into the YOLO-World-v2 framework... improving zero-shot detection on VisDrone by +5.3 mAP'). However, Table 2 shows that this is the combined effect of domain-specific pretraining (8.59→12.2 for L-scale) plus CAGE (12.2→13.9). Controlled comparisons on the same UAV data give CAGE only +0.4 (S), +1.4 (M), and +1.7 (L) mAP. Moreover, on SIMD (Table 3) CAGE slightly decreases L-scale mAP (11.2→11.0). The isolated contribution of the module should be reported in the abstract and conclusion, and the current phrasing should be corrected.
  3. [Section 5.1 / Table 1] The implementation details state that VisDrone images are excluded from pretraining, but Table 1 lists VisDrone as a component of UAVDE-2M with 8,629 images and 457,066 instances. If VisDrone is removed during training, the dataset composition used for the experiments in Table 2 differs from the published UAVDE-2M statistics. Please clarify whether the 'UAV' training data is UAVDE-2M \ VisDrone, and provide the exact instance/image counts after exclusion. If VisDrone images are not fully excluded, the zero-shot evaluation on VisDrone is compromised.
  4. [Section 4.1] The CAGE module is composed of four proposed components: multi-head cross-attention, a gated context refinement head, a global FiLM layer, and a residual path. No ablation isolates the contribution of each component. Given that the module's controlled gains over the YOLO-World-v2 baseline are modest (+0.4 to +1.7 mAP), the necessity and individual benefit of each design choice are not demonstrated. Please add a component-wise ablation study.
minor comments (5)
  1. [Section 4.1] Notation issues: 'D' is used both for depth of text embeddings and is described as 'denotes batch size'; the text embeddings dimension is defined inconsistently. Also, 'Ddonates' should be 'D denotes'. Please clean up the notation.
  2. [Figure 2 caption] Typo: 'V osDrone' should be 'VisDrone'.
  3. [Table 3 / Section 5.2] The text says 'our L-scale model achieves a highly competitive performance comparable to its baseline', but Table 3 shows YOLO-World-v2-L w/ CAGE (11.0) is slightly lower than YOLO-World-v2-L (11.2). Please rephrase to avoid implying a gain where there is none.
  4. [Table 2 caption] The abbreviation 'UAV' is defined in the caption, but the caption does not state that the UAVDE-2M component excludes VisDrone. Please make this explicit for reproducibility.
  5. [General] The paper uses 'UA V' with a space in many places (e.g., 'UA VDE-2M', 'UA V platform'). Please standardize to 'UAV' throughout.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the dataset and CAGE module are validated against external zero-shot benchmarks, and no claimed result reduces by construction to its training inputs.

full rationale

The paper's central claims are empirical: constructing UAVDE-2M/UAVCAP-15K and inserting the CAGE module into YOLO-World-v2 improves zero-shot detection on VisDrone and SIMD. These evaluations use external test sets with class names as text prompts, and the training protocol explicitly excludes VisDrone images ('images from VisDrone[43] are excluded'), so the reported gains are not a re-evaluation of training data. The pseudo-labels produced by Qwen2.5-VL for the COD/FOP subsets are indeed not quantitatively validated, but that is a data-quality/correctness risk, not circularity: the LVLM is an external pretrained model, and the downstream evaluation is on held-out benchmarks. The CAGE module is a new architectural component, not a renamed existing result, and no parameter is fitted to the evaluation sets and then reported as a prediction. There is also no load-bearing self-citation chain or imported uniqueness theorem. The ablation in Table 4 isolates dataset pretraining from architecture, further supporting that the gains are not definitionally forced. Accordingly, no circular step is identifiable, and the appropriate score is 0.

Assumptions & free parameters 3 free parameters · 4 assumptions · 1 invented entities

The central claims rest on the quality of auto-generated labels, the coherence of merged categories, and the validity of the zero-shot protocol. These are unverified domain assumptions.

free parameters (3)
  • Frame redundancy similarity threshold
    Used for discarding redundant frames via DINOv2 cosine similarity in the UAV-Label Engine; value not reported, set by hand. (Section 3.1)
  • Evaluation confidence threshold = 0.001
    Used for zero-shot evaluation on VisDrone, stated in Figure 2 caption; hand-chosen.
  • Training epochs = 100
    Training budget, not fitted but a model choice that affects results. (Section 5.1)
assumptions (4)
  • domain assumption LVLM-generated pseudo-labels (Qwen2.5-VL) are accurate enough for training OVD models.
    Section 3.1 uses Qwen2.5-VL for class prediction and re-classification; no human verification or quality metrics are reported.
  • domain assumption Combining annotations from many source datasets with different label schemas produces a coherent category space.
    Section 3.1 and Table 1 merge 1,853 categories from 27 datasets; label consistency is assumed, not demonstrated.
  • domain assumption Excluding VisDrone from pretraining makes VisDrone evaluation a valid zero-shot test.
    Section 5.1 states exclusion, but the exclusion process is not described and VisDrone appears in Table 1; the assumption underlies the central zero-shot claim.
  • standard math Standard backpropagation and optimizer convergence.
    Training follows Ultralytics defaults; no proof given, but this is standard practice.
invented entities (1)
  • CAGE module
    purpose: Cross-modal fusion of text and visual features via cross-attention, gating, FiLM and residual connection.
    Only supported by the paper's own experiments; no external validation or theoretical justification.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Light-Weight Cross-Modal Enhancement Method with Benchmark Construction for UAV-based Open-Vocabulary Object Detection." pith.science (2026). https://pith.science/paper/ZUJVQMDL

@misc{pith2026250906011,
  author       = {Pith},
  title        = {Pith review of: Light-Weight Cross-Modal Enhancement Method with Benchmark Construction for UAV-based Open-Vocabulary Object Detection},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ZUJVQMDL}},
  note         = {Machine review of arXiv:2509.06011}
}
read the original abstract

Open-Vocabulary Object Detection (OVD) faces severe performance degradation when applied to UAV imagery due to the domain gap from ground-level datasets. To address this challenge, we propose a complete UAV-oriented solution that combines both dataset construction and model innovation. First, we design a refined UAV-Label Engine, which efficiently resolves annotation redundancy, inconsistency, and ambiguity, enabling the generation of largescale UAV datasets. Based on this engine, we construct two new benchmarks: UAVDE-2M, with over 2.4M instances across 1,800+ categories, and UAVCAP-15K, providing rich image-text pairs for vision-language pretraining. Second, we introduce the Cross-Attention Gated Enhancement (CAGE) module, a lightweight dual-path fusion design that integrates cross-attention, adaptive gating, and global FiLM modulation for robust textvision alignment. By embedding CAGE into the YOLO-World-v2 framework, our method achieves significant gains in both accuracy and efficiency, notably improving zero-shot detection on VisDrone by +5.3 mAP while reducing parameters and GFLOPs, and demonstrating strong cross-domain generalization on SIMD. Extensive experiments and real-world UAV deployment confirm the effectiveness and practicality of our proposed solution for UAV-based OVD

Figures

Figures reproduced from arXiv: 2509.06011 by the authors.

Figure 1
Figure 1. Image of our UAV platform. Our UAV experimental platform is equipped with a gimbal camera and an onboard com￾puting device (NVIDIA ORIN NX 16G). troduced open-vocabulary object detection (OVD)[24, 40]. By leveraging vision-language models (VLMs) such as CLIP[29] to align textual and visual information, OVD methods can detect novel categories not present in the train￾ing set, demonstrating remarkable zero-shot capabi… view at source ↗
Figure 3
Figure 3. Common Challenges in Open-Vocabulary Detection for UAV Imagery. Common failure modes for OVD in aerial per￾spectives include false negatives (missed detections) and semantic mismatches. 2. Related Works 2.1. Open-Vocabulary Object Detection OVD signifies a crucial evolution in computer vision, mark￾ing a departure from conventional closed-set paradigms to￾ward a more flexible open-vocabulary framework. Pioneer￾ing r… view at source ↗
Figure 2
Figure 2. Comparison of YOLO-World’s Zero-Shot Perfor￾mance on the COCO and VisDrone Datasets. The green stars demonstrate the mAP in COCO dataset, and the yellow ones demonstrate the mAP in VosDrone dataset. Specifically, for the VisDrone dataset, we use a confidence threshold of 0.001, a set￾ting that is kept consistent across all following experiments. To effectively embed textual guidance into UAV visual representations, … view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Word clouds illustrating the class distribution of the datasets. In contrast to the pre-annotated FOD datasets (e.g., (a), (b)), the COD datasets processed by our LVLM pipeline (e.g., (c), (d)) exhibit a significantly richer and more diverse set of class la￾bels. 3 [P…
Figure 5
Figure 5. Figure 5: Prompt and a sample of UAVCAP-15K. The UAVCAP-15K dataset utilizes a format of long-form, detailed descriptions for its captions. Moreover, it employs prompting techniques to enforce a consistent structure, ensuring all captions describe the image content in a uniform …
Figure 6
Figure 6. Figure 6: Detail of cross-attention gated enhancement module. The A is the overview of cross-attention gated enhancement module, B and C are the detail of gated head and cross attention respectively. visual features from the PAN structure and text embed￾dings from the shared tex…
Figure 7
Figure 7. Figure 7: Visualization of open-vocabulary detection results from a real-world UAV perspective. Our method demonstrates better detection performance. language and vision with a controllable gate, global FiLM modulation and residual bypass. Plugged into YOLO￾World-v2 with no othe…

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

44 extracted references · 39 canonical work pages

  1. [1]

    Qwen2.5-vl technical report.arXiv preprint arXiv:2502.13923, 2025

    Shuai Bai, Keqin Chen, Xuejing Liu, Jialin Wang, Wenbin Ge, Sibo Song, Kai Dang, Peng Wang, Shijie Wang, Jun Tang, Humen Zhong, Yuanzhi Zhu, Mingkun Yang, Zhao- hai Li, Jianqiang Wan, Pengfei Wang, Wei Ding, Zheren Fu, Yiheng Xu, Jiabo Ye, Xi Zhang, Tianbao Xie, Zesen Cheng, Hang Zhang, Zhibo Yang, Haiyang Xu, and Jun- yang Lin. Qwen2.5-vl technical repor...

    arXiv 2025

  2. [2]

    Dataset: Traffic images captured from uavs for use in training machine vision algo- rithms for traffic management.Data, 7(5), 2022

    Sergio Bemposta Rosende, Sergio Ghisler, Javier Fernández- Andrés, and Javier Sánchez-Soriano. Dataset: Traffic images captured from uavs for use in training machine vision algo- rithms for traffic management.Data, 7(5), 2022. 4

    work page 2022

  3. [3]

    Au-air: A multi-modal un- manned aerial vehicle dataset for low altitude traffic surveil- lance

    Ilker Bozcan and Erdal Kayacan. Au-air: A multi-modal un- manned aerial vehicle dataset for low altitude traffic surveil- lance. In2020 IEEE International Conference on Robotics and Automation (ICRA), pages 8504–8510, 2020. 4

    work page 2020

  4. [4]

    Large-scale structure from motion with semantic con- straints of aerial images

    Yu Chen, Yao Wang, Peng Lu, Yisong Chen, and Guoping Wang. Large-scale structure from motion with semantic con- straints of aerial images. InChinese Conference on Pattern Recognition and Computer Vision (PRCV), pages 347–359. Springer, 2018. 4

    2018

  5. [5]

    YOLO-MS: Rethink- ing Multi-Scale Representation Learning for Real-time Ob- ject Detection.IEEE Transactions on Pattern Analysis and Machine Intelligence, pages 1–14, 2025

    Yuming Chen, Xinbin Yuan, Jiabao Wang, Ruiqi Wu, Xiang Li, Qibin Hou, and Ming-Ming Cheng. YOLO-MS: Rethink- ing Multi-Scale Representation Learning for Real-time Ob- ject Detection.IEEE Transactions on Pattern Analysis and Machine Intelligence, pages 1–14, 2025. 3

    work page 2025

  6. [6]

    Ef-detr: A lightweight transformer-based object detector with an encoder-free neck

    Siyi Cheng, Jingnan Song, Mingliang Zhou, Xuekai Wei, Huayan Pu, Jun Luo, and Weijia Jia. Ef-detr: A lightweight transformer-based object detector with an encoder-free neck. IEEE Transactions on Industrial Informatics, 20(11):12994– 13002, 2024. 3

    work page 2024

  7. [7]

    Yolo-world: Real-time open- vocabulary object detection

    Tianheng Cheng, Lin Song, Yixiao Ge, Wenyu Liu, Xing- gang Wang, and Ying Shan. Yolo-world: Real-time open- vocabulary object detection. InProc. IEEE Conf. Computer Vision and Pattern Recognition (CVPR), 2024. 1, 3, 6, 7

    work page 2024

  8. [8]

    Vision-based uav self- positioning in low-altitude urban environments.IEEE Trans- actions on Image Processing, 33:493–508, 2024

    Ming Dai, Enhui Zheng, Zhenhua Feng, Lei Qi, Jiedong Zhuang, and Wankou Yang. Vision-based uav self- positioning in low-altitude urban environments.IEEE Trans- actions on Image Processing, 33:493–508, 2024. 4

    work page 2024

Show all 44 references
  1. [9]

    Counting cattle in uav images using convolutional neural network.Remote Sensing Applications: Society and Environment, 29:100900, 2023

    Fabricio de Lima Weber, Vanessa Aparecida de Moraes We- ber, Pedro Henrique de Moraes, Edson Takashi Matsub- ara, Débora Maria Barroso Paiva, Marina de Nadai Bonin Gomes, Luiz Orcírio Fialho de Oliveira, Sérgio Raposo de Medeiros, and Maria Istela Cagnin. Counting cattle in ua...

    2023

  2. [10]

    Cross-layer feature pyramid transformer for small object detection in aerial images.IEEE Transactions on Geo- science and Remote Sensing, 63:1–14, 2025

    Zewen Du, Zhenjiang Hu, Guiyu Zhao, Ying Jin, and Hong- bin Ma. Cross-layer feature pyramid transformer for small object detection in aerial images.IEEE Transactions on Geo- science and Remote Sensing, 63:1–14, 2025. 3

    2025

  3. [11]

    Llmdet: Learning strong open-vocabulary object detectors under the supervision of large language models

    Shenghao Fu, Qize Yang, Qijie Mo, Junkai Yan, Xihan Wei, Jingke Meng, Xiaohua Xie, and Wei-Shi Zheng. Llmdet: Learning strong open-vocabulary object detectors under the supervision of large language models. InProceedings of the Computer Vision and Pattern Recognition Conferenc...

    2025

  4. [12]

    Lvis: A dataset for large vocabulary instance segmentation

    Agrim Gupta, Piotr Dollar, and Ross Girshick. Lvis: A dataset for large vocabulary instance segmentation. InPro- ceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2019. 1

    2019

  5. [13]

    An ensemble deep learning method with optimized weights for drone-based water rescue and surveillance.Inte- grated Computer-Aided Engineering, pages 1–15, 2021

    Jan G ˛ asienica-Józkowy, Mateusz Knapik, and Boguslaw Cy- ganek. An ensemble deep learning method with optimized weights for drone-based water rescue and surveillance.Inte- grated Computer-Aided Engineering, pages 1–15, 2021. 3, 4

    2021

  6. [14]

    A context-scale- aware detector and a new benchmark for remote sensing small weak object detection in unmanned aerial vehicle im- ages.Int

    Wei Han, Jun Li, Sheng Wang, Yi Wang, Jining Yan, Runyu Fan, Xiaohan Zhang, and Lizhe Wang. A context-scale- aware detector and a new benchmark for remote sensing small weak object detection in unmanned aerial vehicle im- ages.Int. J. Appl. Earth Obs. Geoinformation, 112:102966,

  7. [15]

    Multisized object detection using space- borne optical imagery.IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 13:3032– 3046, 2020

    Muhammad Haroon, Muhammad Shahzad, and Muham- mad Moazam Fraz. Multisized object detection using space- borne optical imagery.IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 13:3032– 3046, 2020. 7

    2020

  8. [16]

    Flame 3 dataset: Unleashing the power of radiometric thermal uav imagery for wildfire man- agement, 2024

    Bryce Hopkins, Leo ONeill, Michael Marinaccio, Eric Row- ell, Russell Parsons, Sarah Flanary, Irtija Nazim, Carl Seiel- stad, and Fatemeh Afghah. Flame 3 dataset: Unleashing the power of radiometric thermal uav imagery for wildfire man- agement, 2024. 4

    2024

  9. [17]

    Meng-Ru Hsieh, Yen-Liang Lin, and Winston H. Hsu. Drone-based object counting by spatially regularized re- gional proposal networks. InThe IEEE International Con- ference on Computer Vision (ICCV). IEEE, 2017. 4

    2017

  10. [18]

    Mdetr - mod- ulated detection for end-to-end multi-modal understanding

    Aishwarya Kamath, Mannat Singh, Yann LeCun, Gabriel Synnaeve, Ishan Misra, and Nicolas Carion. Mdetr - mod- ulated detection for end-to-end multi-modal understanding. In2021 IEEE/CVF International Conference on Computer Vision (ICCV), pages 1760–1770, 2021. 2, 7

    2021

  11. [19]

    Yolov11: An overview of the key architectural enhancements, 2024

    Rahima Khanam and Muhammad Hussain. Yolov11: An overview of the key architectural enhancements, 2024. 6, 7

    2024

  12. [20]

    Autonomous, onboard vision-based trash and litter detection in low altitude aerial images collected by an unmanned aerial vehicle.Remote Sensing, 13(5), 2021

    Marek Kraft, Mateusz Piechocki, Bartosz Ptak, and Krzysztof Walas. Autonomous, onboard vision-based trash and litter detection in low altitude aerial images collected by an unmanned aerial vehicle.Remote Sensing, 13(5), 2021. 4

    2021

  13. [21]

    Lawrence Zitnick, and Piotr Dollár

    Tsung-Yi Lin, Michael Maire, Serge Belongie, Lubomir Bourdev, Ross Girshick, James Hays, Pietro Perona, Deva Ramanan, C. Lawrence Zitnick, and Piotr Dollár. Microsoft coco: Common objects in context, 2015. 1

    2015

  14. [22]

    Focal loss for dense object detection

    Tsung-Yi Lin, Priya Goyal, Ross Girshick, Kaiming He, and Piotr Dollár. Focal loss for dense object detection. In2017 IEEE International Conference on Computer Vision (ICCV), pages 2999–3007, 2017. 3

    2017

  15. [23]

    Esod: Effi- cient small object detection on high-resolution images.IEEE Transactions on Image Processing, 34:183–195, 2025

    Kai Liu, Zhihang Fu, Sheng Jin, Ze Chen, Fan Zhou, Rongxin Jiang, Yaowu Chen, and Jieping Ye. Esod: Effi- cient small object detection on high-resolution images.IEEE Transactions on Image Processing, 34:183–195, 2025. 3

    2025

  16. [24]

    Grounding dino: Marrying dino with grounded pre-training for open-set object detection.arXiv preprint arXiv:2303.05499, 2023

    Shilong Liu, Zhaoyang Zeng, Tianhe Ren, Feng Li, Hao Zhang, Jie Yang, Chunyuan Li, Jianwei Yang, Hang Su, Jun Zhu, et al. Grounding dino: Marrying dino with grounded pre-training for open-set object detection.arXiv preprint arXiv:2303.05499, 2023. 1, 2 9

    2023 arXiv

  17. [25]

    Era: A data set and deep learning benchmark for event recog- nition in aerial videos [software and data sets].IEEE Geo- science and Remote Sensing Magazine, 8(4):125–133, 2020

    Lichao Mou, Yuansheng Hua, Pu Jin, and Xiao Xiang Zhu. Era: A data set and deep learning benchmark for event recog- nition in aerial videos [software and data sets].IEEE Geo- science and Remote Sensing Magazine, 8(4):125–133, 2020. 4

    2020

  18. [26]

    A bench- mark and simulator for uav tracking

    Smith Neil Mueller Matthias and Bernard Ghanem. A bench- mark and simulator for uav tracking. InComputer Vision – ECCV 2016, pages 445–461, 2016. 4

    2016

  19. [27]

    Ensem- ble knowledge transfer for semantic segmentation.2018 IEEE Winter Conference on Applications of Computer Vision (WACV), pages 1499–1508, 2018

    Ishan Nigam, Chen Huang, and Deva Ramanan. Ensem- ble knowledge transfer for semantic segmentation.2018 IEEE Winter Conference on Applications of Computer Vision (WACV), pages 1499–1508, 2018. 4

    2018

  20. [28]

    Locate anything on earth: Advancing open-vocabulary ob- ject detection for remote sensing community

    Jiancheng Pan, Yanxing Liu, Yuqian Fu, Muyuan Ma, Jiahao Li, Danda Pani Paudel, Luc Van Gool, and Xiaomeng Huang. Locate anything on earth: Advancing open-vocabulary ob- ject detection for remote sensing community. InProceed- ings of the AAAI Conference on Artificial Intellige...

    2025

  21. [29]

    Learning transferable visual models from natural language supervision

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, and Ilya Sutskever. Learning transferable visual models from natural language supervision. InProceedings of the...

    2021

  22. [30]

    Floodnet: A high resolution aerial imagery dataset for post flood scene understanding.arXiv preprint arXiv:2012.02951, 2020

    Maryam Rahnemoonfar, Tashnim Chowdhury, Argho Sarkar, Debvrat Varshney, Masoud Yari, and Robin Mur- phy. Floodnet: A high resolution aerial imagery dataset for post flood scene understanding.arXiv preprint arXiv:2012.02951, 2020. 4

    2012 arXiv

  23. [31]

    Rescuenet: A high resolution uav semantic segmen- tation dataset for natural disaster damage assessment.Scien- tific Data, 10(1):913, 2023

    Maryam Rahnemoonfar, Tashnim Chowdhury, and Robin Murphy. Rescuenet: A high resolution uav semantic segmen- tation dataset for natural disaster damage assessment.Scien- tific Data, 10(1):913, 2023. 4

    2023

  24. [32]

    Objects365: A large-scale, high-quality dataset for object detection

    Shuai Shao, Zeming Li, Tianyuan Zhang, Chao Peng, Gang Yu, Jing Li, Xiangyu Zhang, and Jian Sun. Objects365: A large-scale, high-quality dataset for object detection. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 8425–8434, 2019. 2, 7

    2019

  25. [33]

    Ladd: Lacmus drone dataset, 2023

    Mikhail Shuranov, Denis Shurenkov, Dmitry Ruzhit- sky, Victoria Martynova, Ekaterina Bykova, and Georgy Perevozchikov. Ladd: Lacmus drone dataset, 2023. 4

    2023

  26. [34]

    Drone-based rgb-infrared cross-modality vehicle detection via uncertainty-aware learning.IEEE Transactions on Cir- cuits and Systems for Video Technology, pages 1–1, 2022

    Yiming Sun, Bing Cao, Pengfei Zhu, and Qinghua Hu. Drone-based rgb-infrared cross-modality vehicle detection via uncertainty-aware learning.IEEE Transactions on Cir- cuits and Systems for Video Technology, pages 1–1, 2022. 1

    2022

  27. [35]

    Seadronessee: A maritime benchmark for detecting humans in open water

    Leon Amadeus Varga, Benjamin Kiefer, Martin Messmer, and Andreas Zell. Seadronessee: A maritime benchmark for detecting humans in open water. InProceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, pages 2260–2270, 2022. 3, 4

    2022

  28. [36]

    Yoloe: Real-time seeing anything, 2025

    Ao Wang, Lihao Liu, Hui Chen, Zijia Lin, Jungong Han, and Guiguang Ding. Yoloe: Real-time seeing anything, 2025. 1, 6

    2025

  29. [37]

    Dac-sdc: Design automation conference system design contest 2022 dataset, 2022

    Xiaowei Xu, Xinyi Zhang, Bei Yu, Xiaobo Sharon Hu, Christopher Rowen, Jingtong Hu, and Yiyu Shi. Dac-sdc: Design automation conference system design contest 2022 dataset, 2022. 4

    2022

  30. [38]

    More clear, more flexible, more precise: A comprehensive oriented object detection benchmark for uav,

    Kai Ye, Haidi Tang, Bowen Liu, Pingyang Dai, Liujuan Cao, and Rongrong Ji. More clear, more flexible, more precise: A comprehensive oriented object detection benchmark for uav,

  31. [39]

    The unmanned aerial vehicle benchmark: Object detection, tracking and baseline.Int

    Hongyang Yu, Guorong Li, Weigang Zhang, Qingming Huang, Dawei Du, Qi Tian, and Nicu Sebe. The unmanned aerial vehicle benchmark: Object detection, tracking and baseline.Int. J. Comput. Vision, 128(5):1141–1159, 2020. 4

    2020

  32. [40]

    Open-vocabulary object detection using captions

    Alireza Zareian, Kevin Dela Rosa, Derek Hao Hu, and Shih- Fu Chang. Open-vocabulary object detection using captions. In2021 IEEE/CVF Conference on Computer Vision and Pat- tern Recognition (CVPR), pages 14388–14397, 2021. 1, 2

    2021

  33. [41]

    Air-to-air visual detection of micro- uavs: An experimental evaluation of deep learning.IEEE Robotics and Automation Letters, 6(2):1020–1027, 2021

    Ye Zheng, Zhang Chen, Dailin Lv, Zhixing Li, Zhenzhong Lan, and Shiyu Zhao. Air-to-air visual detection of micro- uavs: An experimental evaluation of deep learning.IEEE Robotics and Automation Letters, 6(2):1020–1027, 2021. 3, 4

    2021

  34. [42]

    University- 1652: A multi-view multi-source benchmark for drone- based geo-localization.ACM Multimedia, 2020

    Zhedong Zheng, Yunchao Wei, and Yi Yang. University- 1652: A multi-view multi-source benchmark for drone- based geo-localization.ACM Multimedia, 2020. 4

    2020

  35. [43]

    Detection and tracking meet drones challenge.IEEE Transactions on Pattern Anal- ysis and Machine Intelligence, 44(11):7380–7399, 2021

    Pengfei Zhu, Longyin Wen, Dawei Du, Xiao Bian, Heng Fan, Qinghua Hu, and Haibin Ling. Detection and tracking meet drones challenge.IEEE Transactions on Pattern Anal- ysis and Machine Intelligence, 44(11):7380–7399, 2021. 1, 4, 6, 7, 8

    2021

  36. [44]

    Sues-200: A multi-height multi- scene cross-view image benchmark across drone and satel- lite.IEEE Transactions on Circuits and Systems for Video Technology, 33(9):4825–4839, 2023

    Runzhe Zhu, Ling Yin, Mingze Yang, Fei Wu, Yuncheng Yang, and Wenbo Hu. Sues-200: A multi-height multi- scene cross-view image benchmark across drone and satel- lite.IEEE Transactions on Circuits and Systems for Video Technology, 33(9):4825–4839, 2023. 4 10

    2023

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.