Pith. sign in

REVIEW 4 major objections 4 minor 1 cited by

RF-DETR Object Detection vs YOLOv12 : A Study of Transformer-based and CNN-based Architectures for Single-Class and Multi-Class Greenfruit Detection in Complex Orchard Environments Under Label Ambiguity

T0 review · 4 major / 4 minor · reviewed 2026-08-16 · deepseek-v4-flash

Pith's one-line read RF-DETR achieves the highest mAP@50 for greenfruit detection in single-class and multi-class orchard settings, while YOLOv12N leads the stricter mAP@50:95 in single-class.

desk verdict A useful dataset and an interesting RF-DETR vs YOLOv12 comparison, but the paper's own methods sections contradict each other on training protocols, so the headline fairness claim doesn't hold as written. read the letter →

arxiv 2504.13099 v1 pith:RUYPTNH6 submitted 2025-04-17 cs.CV

classification cs.CV
keywords greenfruitdetectionRF-DETRYOLOv12transformerobjectCNNocclusionhandlinglabelambiguityorchardenvironment
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

This paper compares RF-DETR, a transformer-based detector that combines a DINOv2 backbone with deformable attention, against three YOLOv12 CNN variants for detecting immature green apples in a commercial orchard where fruits blend into foliage and are often partially hidden. On a custom set of 857 images with single-class and occluded/non-occluded multi-class labels, RF-DETR attains the highest mAP@50 in both tasks (0.9464 and 0.8298), while YOLOv12N attains the highest single-class mAP@50:95 (0.7620) and YOLOv12L the highest multi-class mAP@50:95 (0.6622). The paper also reports that RF-DETR converges within a small number of epochs, far sooner than YOLOv12X. The aim is to show that the transformer's global context modeling is better suited to occlusion and camouflage in precision agriculture, with YOLOv12 remaining competitive under stricter overlap thresholds and for speed-sensitive deployment.

What carries the argument

The object that carries the comparison is the architectural pair: RF-DETR-Base, a real-time detection transformer combining a DINOv2 backbone, deformable cross-attention, single-scale features, and NMS-free set prediction, versus YOLOv12, a CNN detector built on R-ELAN blocks, 7x7 separable convolutions, and area attention. The paper uses this pair to attribute detection differences to global context in the transformer versus local feature extraction in the CNN. The measured instruments are the standard detection metrics precision, recall, F1, mAP@50, and mAP@50:95, with mAP@50 serving as the headline comparison and mAP@50:95 as the stricter-overlap check.

What would settle it

Retrain both model families on the same split with identical batch size, optimizer, and epoch count, or a fixed compute budget; if YOLOv12X closes or reverses the mAP@50 gap, the claimed architectural advantage is unsupported. A second check would measure inter-annotator agreement on the occluded/non-occluded labels, since low agreement would weaken the multi-class ranking.

Watch

Extended reading notes

Core claim

On its own terms, the paper's central discovery is that the transformer-based RF-DETR-Base model outperforms all tested YOLOv12 configurations at the 50% IoU threshold for greenfruit detection in a cluttered orchard, with single-class mAP@50 of 0.9464 and multi-class mAP@50 of 0.8298. The best YOLOv12 variant, YOLOv12N, surpasses RF-DETR at the stricter mAP@50:95 metric in single-class detection (0.7620 versus 0.7433), and YOLOv12L does the same in multi-class detection (0.6622 versus 0.6530). Qualitatively, RF-DETR detects heavily occluded and camouflaged fruitlets that YOLOv12 misses, and it classifies occlusion status more accurately in the multi-class case. The paper interprets these results as evidence that global context modeling from transformer attention is the decisive factor in label-ambiguous orchard scenes.

Load-bearing premise

The load-bearing premise is that the models were compared under identical experimental settings; the paper itself reports different training epochs and batch sizes for RF-DETR and YOLOv12, so if those differences matter, the architecture comparison is not clean.

Editorial extensions

If this is right

  • If RF-DETR's margin at mAP@50 holds, accuracy-critical tasks such as robotic thinning and early yield estimation should prefer the transformer model over YOLOv12 on similar orchard data.
  • The fast convergence of RF-DETR means domain-specific agricultural detectors can reach stable performance in tens of epochs, cutting compute costs in repeated orchard surveys.
  • YOLOv12N's lead at mAP@50:95 in single-class detection keeps CNN detectors in play for applications that demand tight box alignment or edge deployment.
  • The multi-class occlusion result implies that transformer detectors may reduce false negatives on partially hidden fruit, which is the main counting error in dense canopies.

Reading between the lines

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

  • The reported mAP gap is not necessarily architectural, because the methods give the two model families different training budgets; a matched-compute rerun would test whether the gap persists.
  • The occlusion labels are inherently subjective, so the multi-class comparison would be stronger with inter-annotator agreement measured; low agreement would mean part of the occlusion-classification gap is label noise.
  • RF-DETR's fast convergence hints that transformer-based detectors could be trained effectively with far fewer annotated orchard images, a testable extension for other fruit crops.
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

4 major / 4 minor

Summary. This paper reports an empirical comparison of RF-DETR-Base (a transformer-based detector) and three YOLOv12 variants (X, L, N; CNN-based) for detecting immature green apples in a commercial orchard. The authors constructed a custom dataset of 857 RGB images with two labeling schemes: a single-class scheme (greenfruit) and a multi-class scheme (occluded vs. non-occluded greenfruit). They report precision, recall, F1-score, mAP@50, and mAP@50:95 for each model and task, and they analyze training convergence. The headline results are that RF-DETR achieves the highest mAP@50 in both single-class (0.9464) and multi-class (0.8298) detection, while YOLOv12N achieves the highest single-class mAP@50:95 (0.7620) and YOLOv12L the highest multi-class mAP@50:95 (0.6622). The authors conclude that RF-DETR excels at localizing occluded and camouflaged fruit, while YOLOv12 offers advantages in speed-oriented deployments. The paper also includes qualitative examples and a discussion of label ambiguity.

Significance. If the reported results were obtained under truly identical training protocols and were statistically stable, the paper would provide a useful apples-to-apples benchmark of two popular detection paradigms in a challenging precision-agriculture scenario, with a new field dataset and a practical occlusion-labeling scheme. The qualitative observations about RF-DETR handling occlusion and camouflage are plausible and align with the architectural differences between DETR-style and YOLO-style detectors. However, the paper currently ships no code, no data, no dataset splits, and no uncertainty estimates, and the methods sections contradict each other on the training protocols. As a result, the central comparative claim is not yet substantiated. The work has potential value as an application study, but its current form is not a reliable basis for architecture-level conclusions.

major comments (4)
  1. [Section 2.3.1 vs. Section 2.4] The training protocols are internally inconsistent. Section 2.3.1 states that RF-DETR was trained with a batch size of 8 for 300 epochs, while Section 2.4 states that RF-DETR was trained for 50 epochs on the single-class dataset and 100 epochs on the multi-class dataset with a batch size of approximately 16, and that all YOLOv12 models were trained for 100 epochs. These descriptions cannot both be true. Since Section 2.4 and Section 2.5 claim that both models were trained under 'identical experimental settings,' the contradiction directly undermines the fairness premise of the entire comparison. The authors must clarify the exact training budget (epochs, batch size, optimizer steps, learning rate schedule) actually used for each model and dataset, or the headline mAP differences cannot be attributed to architecture.
  2. [Section 2.5 and Table 1] All performance metrics are reported as single-run point estimates with no error bars, confidence intervals, or repeated-seed variance. The paper treats differences as small as 0.01 (e.g., multi-class mAP@50:95 of 0.6609 for YOLOv12X vs. 0.6530 for RF-DETR, Section 3.2) as meaningful. Without variance estimates or a significance test, these differences are not distinguishable from noise, especially given the small dataset of 857 images and the label-ambiguity issues the authors themselves describe. The authors should provide at least three repeated runs per model and report mean and standard deviation, or justify why a single run is sufficient.
  3. [Section 2.2] The dataset construction section gives no train/validation/test split information. The authors report 857 images and 4,125 object labels for the single-class scheme but do not state how many images were used for training, validation, and testing, whether split was performed at the image level, or whether any images from the same orchard row or session could leak between splits in a way that inflates metrics. Without this information, the reported mAP numbers cannot be interpreted, and the comparison between models may be affected by split-specific variance. This is a load-bearing omission for a benchmark paper.
  4. [Section 1.2 and Section 2.3.1] The claim that RF-DETR 'is the only model to surpass 60% mAP on the COCO dataset to date' (Section 1.2, restated in Section 2.3.1) is factually incorrect and is even contradicted by the same paper, which states that YOLOv12X achieves 68.9 mAP on COCO (Section 2.3.2). This overstatement should be removed or replaced with an accurate citation-specific claim about transformer-based real-time detectors, if that is what is intended.
minor comments (4)
  1. [Section 3.3] The convergence analysis compares RF-DETR (trained for 50 epochs in single-class) with YOLOv12X (trained for 100 epochs). Presenting RF-DETR as 'plateauing within 10 epochs' while YOLOv12 continues to train is misleading when the training budgets differ; the section should at least acknowledge this asymmetry or use an equal-iteration comparison.
  2. [Section 1.2 and Index Terms] The index terms and abstract mention YOLOv13, YOLOv14, YOLOv15, YOLOE, and YOLO World, none of which are evaluated or discussed in the paper. These terms should be removed or properly cited if they are meant to be related work.
  3. [Throughout] There are numerous typographical and formatting errors, including 'comparision' (Section 3.2 caption), 'e ffective' in several places, and inconsistent spacing in equations and references. A thorough copyedit is needed.
  4. [Section 2.2] The label ambiguity discussion is qualitative and does not quantify inter-annotator agreement. Since the paper's multi-class task hinges on the distinction between occluded and non-occluded fruit, reporting a metric such as Cohen's kappa or a per-image ambiguity rate would strengthen the reliability of the ground truth.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the paper is an empirical benchmark whose mAP values are measured against held-out labels, not derived from fitted parameters or self-citations.

full rationale

This paper reports a direct experimental comparison of RF-DETR and YOLOv12 on orchard greenfruit detection. The central claims—RF-DETR mAP@50 of 0.9464 (single-class) and 0.8298 (multi-class)—are evaluation metrics computed from model predictions against ground-truth annotations on a held-out test split. There is no analytical derivation chain in which an output is defined in terms of an input, no fitted parameter that is later relabeled as a prediction, and no uniqueness theorem or architectural ansatz imported from the authors' prior work to force the conclusion. The paper's own references to earlier YOLO comparisons by the same group are contextual and not load-bearing for the reported numbers. The most notable concern is a methodological inconsistency, not circularity: Section 2.3.1 states RF-DETR was trained with batch size 8 for 300 epochs, while Section 2.4 states RF-DETR was trained for 50 epochs (single-class) and 100 epochs (multi-class) with batch size approximately 16, and YOLOv12 models for 100 epochs; Section 2.5 reiterates that both models used the 'same datasets, number of training epochs, learning rates, optimizers, and batch sizes.' If the differing descriptions reflect the actual runs, the comparison may be unfair in training budget, but this affects validity of the empirical conclusion, not circularity: the mAP numbers are still computed from ground truth rather than being equivalent to their inputs by construction. No circular step can be exhibited, so the appropriate score is 0.

Assumptions & free parameters 2 free parameters · 4 assumptions · 0 invented entities

The central claim is empirical and rests on dataset representativeness, annotation reliability, pretrained initialization, and comparability of training protocols. The paper introduces no theoretical objects or fitted constants beyond training hyperparameters chosen by the authors; the main uncharged inputs are the dataset itself (§2.1-2.2), transfer from pretrained weights (§2.3), and single-run metrics treated as decisive (Table 1).

free parameters (2)
  • RF-DETR training epochs = 50 (single-class), 100 (multi-class) in §2.4; 300 stated in §2.3.1
    Training budget differs from YOLOv12's 100 epochs and is internally inconsistent, so epoch count is a hand-chosen factor that confounds the architecture comparison.
  • Batch size = 8 for RF-DETR (§2.3.1); approximately 16 as the stated general setting (§2.4)
    Chosen by the authors; the mismatch undermines the claim of identical training protocols and could affect optimization quality.
assumptions (4)
  • domain assumption The 857-image custom dataset and its manual annotations are sufficient and reliable for measuring greenfruit detection performance.
    The paper acknowledges label ambiguity in §2.2 but reports no inter-annotator agreement or quality control, so the ground truth itself is unverified.
  • domain assumption Pretrained weights from COCO and DINOv2 transfer to orchard imagery in a way that allows an architecture comparison.
    RF-DETR uses a DINOv2-pretrained backbone and YOLOv12 uses pretrained weights; differing pretraining is not controlled (§2.3).
  • ad hoc to paper Single-run metric differences of 0.01 to 0.02 are treated as meaningful without variance estimates.
    The paper draws winner and loser conclusions from Table 1 and Figure 7 with no repeated runs, standard deviations, or significance tests.
  • standard math The mAP@50 and mAP@50:95 formulas in Equations (5) and (6) are the standard COCO definitions.
    These are conventional metrics; the formulas are stated but no evaluation code is provided to verify their implementation.

how reviews work

0 comments
Cite this review

Pith. "Pith review of RF-DETR Object Detection vs YOLOv12 : A Study of Transformer-based and CNN-based Architectures for Single-Class and Multi-Class Greenfruit Detection in Complex Orchard Environments Under Label Ambiguity." pith.science (2026). https://pith.science/paper/RUYPTNH6

@misc{pith2026250413099,
  author       = {Pith},
  title        = {Pith review of: RF-DETR Object Detection vs YOLOv12 : A Study of Transformer-based and CNN-based Architectures for Single-Class and Multi-Class Greenfruit Detection in Complex Orchard Environments Under Label Ambiguity},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/RUYPTNH6}},
  note         = {Machine review of arXiv:2504.13099}
}
read the original abstract

This study conducts a detailed comparison of RF-DETR object detection base model and YOLOv12 object detection model configurations for detecting greenfruits in a complex orchard environment marked by label ambiguity, occlusions, and background blending. A custom dataset was developed featuring both single-class (greenfruit) and multi-class (occluded and non-occluded greenfruits) annotations to assess model performance under dynamic real-world conditions. RF-DETR object detection model, utilizing a DINOv2 backbone and deformable attention, excelled in global context modeling, effectively identifying partially occluded or ambiguous greenfruits. In contrast, YOLOv12 leveraged CNN-based attention for enhanced local feature extraction, optimizing it for computational efficiency and edge deployment. RF-DETR achieved the highest mean Average Precision (mAP50) of 0.9464 in single-class detection, proving its superior ability to localize greenfruits in cluttered scenes. Although YOLOv12N recorded the highest mAP@50:95 of 0.7620, RF-DETR consistently outperformed in complex spatial scenarios. For multi-class detection, RF-DETR led with an mAP@50 of 0.8298, showing its capability to differentiate between occluded and non-occluded fruits, while YOLOv12L scored highest in mAP@50:95 with 0.6622, indicating better classification in detailed occlusion contexts. Training dynamics analysis highlighted RF-DETR's swift convergence, particularly in single-class settings where it plateaued within 10 epochs, demonstrating the efficiency of transformer-based architectures in adapting to dynamic visual data. These findings validate RF-DETR's effectiveness for precision agricultural applications, with YOLOv12 suited for fast-response scenarios. >Index Terms: RF-DETR object detection, YOLOv12, YOLOv13, YOLOv14, YOLOv15, YOLOE, YOLO World, YOLO, You Only Look Once, Roboflow, Detection Transformers, CNNs

Figures

Figures reproduced from arXiv: 2504.13099 by the authors.

Figure 1
Figure 1. Classification of object detection methodologies: Top fea￾tures state-of-the-art CNN-based and Transformer-based methods, widely adopted; Vision Language Models are emerging. Also includes Hybrid, Sparse Coding, and Traditional Feature-based approaches. Preprint submitted to Elsevier April 18, 2025 arXiv:2504.13099v1 [cs.CV] 17 Apr 2025 [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. CNN vs Transformer-based model performance comparison fo￾cusing on YOLOv12 (CNN-based) and RF-DETR (Transformer-based) architectures: (a) RF-DETR object detection model benchmark evalu￾ation with YOLO11, YOLOv8 and other DETR-based object detection models ; (b)RF-DETR evaluation on the RF100-VL dataset, highlight￾ing domain adaptability and edge deployment potential. ; and (c) Per￾formance overview of recent CNN-bas… view at source ↗
Figure 3
Figure 3. Overview of data collection setup and environment: a) Flow diagram showing the methodology of RF-DETR vs YOLOv12 comparision ; b) Map highlighting the study location in Prosser, Washington, USA ; c) of ’Scifresh’ apple trees, known as Jazz apples; d) The robotic platform used for image acquisition, featuring an Intel RGB-D camera mounted on a UR5e robotic arm, capturing images of immature greenfruits in complex orch… view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: (a) RF-DETR Architecture diagram for object detection ; (b) YOLOv12 Architecture Diagram for object detection transformer-based detectors. The neck architecture integrates FlashAttention-optimized area attention, dividing feature maps into four horizontal/vertical regi…
Figure 5
Figure 5. Figure 5: Visual comparison of single-class greenfruit detection using RF-DETR and YOLOv12 in complex orchard scenes. a) Three clustered greenfruits partially occluded by dense canopy; RF-DETR detected all, YOLOv12 missed one. b) A camouflaged greenfruit blending into the canopy…
Figure 6
Figure 6. Figure 6: Visual comparison of multi-class greenfruit detection by RF-DETR and YOLOv12 under label ambiguity. (a) A dense fruit cluster at the image edge; YOLOv12 over-detected with false positives, while RF-DETR correctly detected 5 true greenfruits. (b) An occluded apple at th…
Figure 7
Figure 7. Figure 7: Mean Average Precision (mAP) comparison for greenfruit detection using RF-DETR and YOLOv12 object detection models: a) mAP@50 for single-class detection. b) mAP@50 and mAP@50:95 for multi-class detection and multi-class settings is emblematic of the inherent strengths …
Figure 8
Figure 8. Figure 8: Training Dynamics and Model Convergence Analysis: mAP@50 vs. Epoch Curves for Object Detection Models. (a) Single-class greenfruit detection showing the performance trajectory of RF-DETR and YOLOv12X models over training epochs. (b) Multi-class greenfruit detection com…

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Container damage detection using advanced computer vision model Yolov12 vs Yolov11 vs RF-DETR A comparative analysis

    cs.CV 2025-06 reject novelty 4.0 of 10

    A comparative benchmark of YOLOv11, YOLOv12 and RF-DETR for container damage detection shows mixed results, with RF-DETR favored only on three hand-picked hard examples.

Reference graph

Works this paper leans on

55 extracted references · 38 canonical work pages · cited by 1 Pith paper

  1. [1]

    Masmoudi, H

    M. Masmoudi, H. Ghazzai, M. Frikha, Y . Massoud, Object detection learning techniques for autonomous vehicle applications, in: 2019 IEEE international conference on vehicular electronics and safety (ICVES), IEEE, 2019, pp. 1–5

  2. [2]

    Hnewa, H

    M. Hnewa, H. Radha, Object detection under rainy conditions for au- tonomous vehicles: A review of state-of-the-art and emerging techniques, IEEE Signal Processing Magazine 38 (2020) 53–67

  3. [3]

    Elakkiya, V

    R. Elakkiya, V . Subramaniyaswamy, V . Vijayakumar, A. Mahanti, Cer- vical cancer diagnostics healthcare system using hybrid object detection adversarial networks, IEEE Journal of Biomedical and Health Informatics 26 (2021) 1464–1471

  4. [4]

    P. K. Mishra, G. Saroha, A study on video surveillance system for object detection and tracking, in: 2016 3rd international conference on com- puting for sustainable global development (INDIACom), IEEE, 2016, pp. 221–226

  5. [5]

    C. M. Badgujar, A. Poulose, H. Gan, Agricultural object detection with you only look once (yolo) algorithm: A bibliometric and systematic lit- erature review, Computers and Electronics in Agriculture 223 (2024) 109090

  6. [6]

    I. Sa, Z. Ge, F. Dayoub, B. Upcroft, T. Perez, C. McCool, Deepfruits: A fruit detection system using deep neural networks, sensors 16 (2016) 1222

  7. [7]

    Singh, R

    P. Singh, R. Krishnamurthi, Iot-based real-time object detection system for crop protection and agriculture field security, Journal of Real-Time Image Processing 21 (2024) 106

  8. [8]

    L. Yang, T. Noguchi, Y . Hoshino, Development of a pumpkin fruits pick- and-place robot using an rgb-d camera and a yolo based object detection ai model, Computers and Electronics in Agriculture 227 (2024) 109625

Show all 55 references
  1. [9]

    J. Gu, Z. Wang, J. Kuen, L. Ma, A. Shahroudy, B. Shuai, T. Liu, X. Wang, G. Wang, J. Cai, et al., Recent advances in convolutional neural networks, Pattern recognition 77 (2018) 354–377

  2. [10]

    Nimma, Z

    D. Nimma, Z. Zhou, Intelpvt: intelligent patch-based pyramid vision transformers for object detection and classification, International Journal of Machine Learning and Cybernetics 15 (2024) 1767–1778

  3. [11]

    H. Liu, Y . Zhan, J. Sun, Q. Mao, T. Wu, A transformer-based model with feature compensation and local information enhancement for end-to- end pest detection, Computers and Electronics in Agriculture 231 (2025) 109920

  4. [12]

    Y . Zang, W. Li, J. Han, K. Zhou, C. C. Loy, Contextual object detection with multimodal large language models, International Journal of Com- puter Vision 133 (2025) 825–843. 14

  5. [13]

    S. Fu, Q. Yang, Q. Mo, J. Yan, X. Wei, J. Meng, X. Xie, W.-S. Zheng, Llmdet: Learning strong open-vocabulary object detectors under the su- pervision of large language models, arXiv preprint arXiv:2501.18954 (2025)

  6. [14]

    C.-Y . Fu, M. Shvets, A. C. Berg, Retinamask: Learning to predict masks improves state-of-the-art single-shot detection for free, arXiv preprint arXiv:1901.03353 (2019)

  7. [15]

    M. Tan, R. Pang, Q. V . Le, E fficientdet: Scalable and e fficient object detection, in: Proceedings of the IEEE /CVF conference on computer vision and pattern recognition, 2020, pp. 10781–10790

  8. [16]

    Redmon, S

    J. Redmon, S. Divvala, R. Girshick, A. Farhadi, You only look once: Uni- fied, real-time object detection, in: Proceedings of the IEEE conference on computer vision and pattern recognition, 2016, pp. 779–788

  9. [17]

    Sapkota, Z

    R. Sapkota, Z. Meng, M. Churuvija, X. Du, Z. Ma, M. Karkee, Compre- hensive performance evaluation of yolo11, yolov10, yolov9 and yolov8 on detecting and counting fruitlet in complex orchard environments, arXiv preprint arXiv:2407.12040 (2024)

  10. [18]

    K. He, G. Gkioxari, P. Doll ´ar, R. Girshick, Mask r-cnn, in: Proceedings of the IEEE international conference on computer vision, 2017, pp. 2961– 2969

  11. [19]

    X. Dai, Y . Chen, J. Yang, P. Zhang, L. Yuan, L. Zhang, Dynamic detr: End-to-end object detection with dynamic attention, in: Proceedings of the IEEE /CVF international conference on computer vision, 2021, pp. 2988–2997

  12. [20]

    X. Zhu, W. Su, L. Lu, B. Li, X. Wang, J. Dai, Deformable detr: De- formable transformers for end-to-end object detection, arXiv preprint arXiv:2010.04159 (2020)

  13. [21]

    Hosang, R

    J. Hosang, R. Benenson, B. Schiele, Learning non-maximum suppres- sion, in: Proceedings of the IEEE conference on computer vision and pattern recognition, 2017, pp. 4507–4515

  14. [22]

    Carion, F

    N. Carion, F. Massa, G. Synnaeve, N. Usunier, A. Kirillov, S. Zagoruyko, End-to-end object detection with transformers, in: European conference on computer vision, Springer, 2020, pp. 213–229

  15. [23]

    X. Ren, D. Ramanan, Histograms of sparse codes for object detection, in: Proceedings of the IEEE conference on computer vision and pattern recognition, 2013, pp. 3246–3253

  16. [24]

    Y . Xie, W. Zhang, C. Li, S. Lin, Y . Qu, Y . Zhang, Discriminative object tracking via sparse representation and online dictionary learning, IEEE transactions on cybernetics 44 (2013) 539–553

  17. [25]

    O’shea, R

    K. O’shea, R. Nash, An introduction to convolutional neural networks, arXiv preprint arXiv:1511.08458 (2015)

  18. [26]

    Soydaner, Attention mechanism in neural networks: where it comes and where it goes, Neural Computing and Applications 34 (2022) 13371– 13385

    D. Soydaner, Attention mechanism in neural networks: where it comes and where it goes, Neural Computing and Applications 34 (2022) 13371– 13385

  19. [27]

    A. Khan, Z. Rauf, A. Sohail, A. R. Khan, H. Asif, A. Asif, U. Farooq, A survey of the vision transformers and their cnn-transformer based vari- ants, Artificial Intelligence Review 56 (2023) 2917–2970

  20. [28]

    Alzubaidi, J

    L. Alzubaidi, J. Zhang, A. J. Humaidi, A. Al-Dujaili, Y . Duan, O. Al- Shamma, J. Santamar´ıa, M. A. Fadhel, M. Al-Amidie, L. Farhan, Review of deep learning: concepts, cnn architectures, challenges, applications, future directions, Journal of big Data 8 (2021) 1–74

  21. [29]

    S. Chen, Y . Liu, X. Gao, Z. Han, Mobilefacenets: Efficient cnns for accu- rate real-time face verification on mobile devices, in: Chinese conference on biometric recognition, Springer, 2018, pp. 428–438

  22. [30]

    Girshick, J

    R. Girshick, J. Donahue, T. Darrell, J. Malik, Rich feature hierarchies for accurate object detection and semantic segmentation, in: Proceedings of the IEEE conference on computer vision and pattern recognition, 2014, pp. 580–587

  23. [31]

    Sapkota, M

    R. Sapkota, M. Karkee, Comparing yolov11 and yolov8 for instance seg- mentation of occluded and non-occluded immature green fruits in com- plex orchard environment, arXiv preprint arXiv:2410.19869 (2024)

  24. [32]

    Y . Tian, Q. Ye, D. Doermann, Yolov12: Attention-centric real-time object detectors, arXiv preprint arXiv:2502.12524 (2025)

  25. [33]

    Sapkota, R

    R. Sapkota, R. Qureshi, M. F. Calero, C. Badjugar, U. Nepal, A. Poulose, P. Zeno, U. B. P. Vaddevolu, S. Khan, M. Shoman, et al., Yolov10 to its genesis: a decadal and comprehensive review of the you only look once (yolo) series, arXiv preprint arXiv:2406.19407 (2024)

  26. [34]

    Sapkota, M

    R. Sapkota, M. Karkee, Improved yolov12 with llm-generated synthetic data for enhanced apple detection and benchmarking against yolov11 and yolov10, arXiv preprint arXiv:2503.00057 (2025)

  27. [35]

    Z. Meng, X. Du, R. Sapkota, Z. Ma, H. Cheng, Yolov10-pose and yolov9- pose: Real-time strawberry stalk pose detection models, Computers in Industry 165 (2025) 104231

  28. [36]

    W. Liu, D. Anguelov, D. Erhan, C. Szegedy, S. Reed, C.-Y . Fu, A. C. Berg, Ssd: Single shot multibox detector, in: Computer Vision–ECCV 2016: 14th European Conference, Amsterdam, The Netherlands, October 11–14, 2016, Proceedings, Part I 14, Springer, 2016, pp. 21–37

  29. [37]

    T.-Y . Lin, P. Goyal, R. Girshick, K. He, P. Doll ´ar, Focal loss for dense object detection, in: Proceedings of the IEEE international conference on computer vision, 2017, pp. 2980–2988

  30. [38]

    D. Wang, Z. Li, X. Du, Z. Ma, X. Liu, Farmland obstacle detection from the perspective of uavs based on non-local deformable detr, Agriculture 12 (2022) 1983

  31. [39]

    H. Lin, J. Liu, X. Li, L. Wei, Y . Liu, B. Han, Z. Wu, Dcea: Detr with concentrated deformable attention for end-to-end ship detection in sar im- ages, IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing (2024)

  32. [40]

    Z. Zong, G. Song, Y . Liu, Detrs with collaborative hybrid assignments training, in: Proceedings of the IEEE /CVF international conference on computer vision, 2023, pp. 6748–6758

  33. [41]

    Zhang, Y

    Y . Zhang, Y . Wu, H. Xu, Y . Xie, Y . Zhang, Improved co-detr with dropkey and its application to hot work detection, Concurrency and Computation: Practice and Experience 37 (2025) e70020

  34. [42]

    Y . Fang, B. Liao, X. Wang, J. Fang, J. Qi, R. Wu, J. Niu, W. Liu, You only look at one sequence: Rethinking transformer in vision through object detection, Advances in Neural Information Processing Systems 34 (2021) 26183–26197

  35. [43]

    Y . Zhao, W. Lv, S. Xu, J. Wei, G. Wang, Q. Dang, Y . Liu, J. Chen, Detrs beat yolos on real-time object detection, in: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2024, pp. 16965–16974

  36. [44]

    Minderer, A

    M. Minderer, A. Gritsenko, A. Stone, M. Neumann, D. Weissenborn, A. Dosovitskiy, A. Mahendran, A. Arnab, M. Dehghani, Z. Shen, et al., Simple open-vocabulary object detection, in: European conference on computer vision, Springer, 2022, pp. 728–755

  37. [45]

    Heigold, M

    G. Heigold, M. Minderer, A. Gritsenko, A. Bewley, D. Keysers, M. Luˇci´c, F. Yu, T. Kipf, Video owl-vit: Temporally-consistent open-world localiza- tion in video, in: Proceedings of the IEEE/CVF International Conference on Computer Vision, 2023, pp. 13802–13811

  38. [46]

    B. Wang, K. Huang, B. Li, Y . Yan, L. Zhang, H. Lu, Y . He, E ffowt: Transfer visual language models to open-world tracking e fficiently and effectively, arXiv preprint arXiv:2504.05141 (2025)

  39. [47]

    Zhang, F

    H. Zhang, F. Li, S. Liu, L. Zhang, H. Su, J. Zhu, L. M. Ni, H.-Y . Shum, Dino: Detr with improved denoising anchor boxes for end-to-end object detection, arXiv preprint arXiv:2203.03605 (2022)

  40. [48]

    Robicheaux, M

    P. Robicheaux, M. Popov, A. Madan, I. Robinson, J. Nelson, D. Ramanan, N. Peri, Roboflow100-vl: A multi-domain object detection benchmark for vision-language models, Roboflow (2025)

  41. [49]

    Sapkota, A

    R. Sapkota, A. Paudel, M. Karkee, Zero-shot automatic annotation and instance segmentation using llm-generated datasets: Eliminating field imaging and manual annotation for deep learning model development, arXiv preprint arXiv:2411.11285 (2024)

  42. [50]

    Sapkota, D

    R. Sapkota, D. Ahmed, M. Churuvija, M. Karkee, Immature green ap- ple detection and sizing in commercial orchards using yolov8 and shape fitting techniques, IEEE Access 12 (2024) 43436–43452

  43. [51]

    Sapkota, M

    R. Sapkota, M. Karkee, Yolo11 and vision transformers based 3d pose estimation of immature green fruits in commercial apple orchards for robotic thinning, arXiv preprint arXiv:2410.19846 (2024)

  44. [52]

    Q. Liu, H. Meng, R. Zhao, X. Ma, T. Zhang, W. Jia, Green apple detector based on optimized deformable detection transformer, Agriculture 15 (2024) 75

  45. [53]

    Sapkota, S

    R. Sapkota, S. Raza, M. Shoman, A. Paudel, M. Karkee, Multimodal large language models for image, text, and speech data augmentation: A survey, arXiv preprint arXiv:2501.18648 (2025)

  46. [54]

    Q. Liu, J. Lv, C. Zhang, Mae-yolov8-based small object detection of green crisp plum in real complex orchard environments, Computers and Electronics in Agriculture 226 (2024) 109458

  47. [55]

    J. Lv, Z. Wu, P. Zhou, J. Huang, G. Liu, Y . Gu, H. Rong, L. Zou, Fcae- yolov8n: a target detection method for immature grape clusters, New Zealand Journal of Crop and Horticultural Science (2024) 1–19. 15

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.