Pith. sign in

REVIEW 4 major objections 6 minor 1 cited by

Enhancing Large Vision Model in Street Scene Semantic Understanding through Leveraging Posterior Optimization Trajectory

T0 review · 4 major / 6 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read This paper claims that a frozen iGPT large vision model with a small ASSP head, trained through a Posterior Optimization Trajectory-guided scheme, reaches 99.99 mIoU on Cityscapes and 82.06 mIoU on CamVid while converging within 10 epochs.

desk verdict Interesting training-time logit-refinement idea, but the 99.99 mIoU claim is unsupported because the paper never specifies whether label-dependent POTGen runs at inference. read the letter →

arxiv 2501.01710 v2 pith:BY4T3NKE submitted 2025-01-03 cs.CV cs.LGcs.RO

Wei-Bin Kou , Qingfeng Lin , Ming Tang , Jingreng Lei , Shuai Wang , Rongguang Ye , Guangxu Zhu , Yik-Chung Wu This is my paper · ORCID
classification cs.CVcs.LGcs.RO
keywords semanticsegmentationautonomousdrivinglargevisionmodelsiGPTposterioroptimizationtrajectorytrainingaccelerationstreetsceneunderstandingPOT-guided
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 tries to establish that the under-fitting a vehicle's semantic-segmentation model faces as onboard data accumulates can be avoided by keeping a pretrained large vision model (iGPT) frozen and training only a small downstream head, and that the cost of training that head can be slashed by a Posterior Optimization Trajectory (POT)-guided scheme. The central experimental claim is that the resulting LVM+POTGui system reaches 99.99 mIoU on Cityscapes and 82.06 mIoU on CamVid, outperforming the plain LVM baseline by 118.27% and 66.48% respectively, and converges within 10 epochs, about 6 times faster (15 times on Cityscapes). If true, this would mean a near-perfect street-scene segmenter can be obtained by training only a small head on frozen large-model features, a practical recipe for onboard autonomous-driving updates. The method works by unrolling the cross-entropy training loss into K synthetic gradient-descent layers, generating 'future' logits from the ground truth, and blending those logits into the current loss to guide optimization.

What carries the argument

The central object is the Posterior Optimization Trajectory (POT) Generator, a learned unrolling of the cross-entropy loss into K gradient-descent 'layers' that produce future logits O_POT. It defines per-layer updates O_k = O_{k-1} - eta_k alpha_k grad_O LCE(P_Y, O_{k-1}) with learnable step sizes alpha_k and eta_k, and a closed-form gradient (Eq. 11) equal to the averaged difference between predicted and ground-truth class probabilities. The POT-guided optimization (POTGui) loss then replaces the raw logits with a weighted blend sigma O + (1-sigma) O_POT before backpropagating (Eq. 14). This carries the argument by injecting information about K future optimization steps into the current gradient, which the paper argues both accelerates convergence and regularizes the learning.

What would settle it

Take the trained LVM+POTGui model and run a pure forward pass on Cityscapes validation with the POTGen label-guided refinement disabled; if mIoU falls far below the reported 99.99 (e.g., toward the plain LVM's 45.81), then the headline metric depended on ground-truth information injected during evaluation. Conversely, if 99.99 persists without any ground-truth input, the result stands as a genuine near-saturation segmentation score.

Watch

Extended reading notes

Core claim

The paper's discovery claim is that posterior optimization trajectories make a frozen LVM sufficient for street-scene semantic understanding. Concretely, using hidden features from the middle four layers of a pretrained iGPT (zero-shot, parameters frozen) and an ASSP head, the POTGui training scheme reaches 99.99 mIoU/mF1/mPrecision/mRecall on Cityscapes validation and 82.06 mIoU on CamVid test, with class-wise IoU above 80% even for slender classes like Pole and Fence that all compared baselines miss entirely. The mechanism: POTGen unfolds one optimization step into K layers, computes the gradient of cross-entropy w.r.t. the logits at each layer (Eq. 11), performs K learnable gradient-descent updates (Eq. 4), and the final refined logits O_POT (Eq. 12) are blended with the original logits as O_POTGui = sigma O + (1-sigma) O_POT (Eq. 13); the head is then trained against the cross-entropy of these blended logits (Eq. 14). The paper reports that more POTGen layers improve both accuracy and convergence speed up to saturation around 50 layers, and that the method stays accurate and low-variance in real-driving tests on ApolloScapes and CARLA across fog, rain, cloud, and dark conditions.

Load-bearing premise

The load-bearing assumption is that the validation numbers in Table II were computed with the label-guided POTGen refinement switched off at inference; the paper never states whether the ground-truth-dependent K-step logit updates used in the training loss are active during evaluation, and the reported near-perfect scores are only meaningful under that unstated protocol.

Editorial extensions

If this is right

  • On the paper's numbers, a frozen iGPT backbone plus an ASSP head reaches 99.99 mIoU on Cityscapes and 82.06 mIoU on CamVid, making the LVM+POTGui combination the top row in every metric on both datasets.
  • POTGui converges in at most 10 epochs: about 6 times faster than the LVM baseline on CamVid and 15 times faster on Cityscapes (first epoch vs 16th epoch), which is the claimed computational relief for onboard training.
  • Rare and slender classes on CamVid (Pole, Fence, Bicyclist) jump from near 0 IoU in the baselines to 85.27, 93.24, and 91.59 IoU respectively, a class-level gain the paper attributes to the LVM's pretraining plus POTGui's optimization.
  • Increasing the number of POTGen layers improves performance and convergence speed up to a saturation point around 50 layers, after which additional layers add nothing (Fig. 5).
  • In real-driving tests on ApolloScapes and CARLA_ADV across foggy, rainy, cloudy, and dark conditions, LVM+POTGui outperforms DeepLabv3+ on all metrics with smaller frame-to-frame variance.

Reading between the lines

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

  • If the reported Cityscapes mIoU survives a no-label inference check, the practical consequence goes beyond this paper: onboard AD updates could skip backbone fine-tuning entirely and retrain only a small head on frozen LVM features for each new data batch, drastically cutting compute and communication.
  • POTGui is a generic training accelerator: the same unrolled-gradient 'future logits' trick could be applied to detection heads, depth heads, or federated-learning client updates, wherever the loss is differentiable and ground-truth labels are available during training; whether the speedups transfer beyond segmentation is untested here.
  • The comparison set stops at older real-time baselines and the LVM-without-POTGui control; a natural next test is to pit this recipe against modern segmentation transformers and domain-adaptation methods on challenging domain-shift benchmarks, and to report class-wise results on Cityscapes with the same protocol.
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 / 6 minor

Summary. The paper proposes a street-scene semantic segmentation system built from a frozen iGPT backbone plus an ASSP head, and a training-time acceleration scheme called POTGui. POTGen unrolls K gradient-descent steps on the predicted logits toward the ground-truth one-hot labels, then mixes the refined logits with the original logits to form the cross-entropy training target. The authors report 99.99 mIoU on Cityscapes and 82.06 mIoU on CamVid with convergence within a few epochs, and claim a 66.48% improvement and over 6x faster convergence relative to the compared baselines. The paper includes layer-selection experiments, an ablation on the number of unrolled layers, and qualitative real-driving tests.

Significance. The proposed training-time lookahead idea—unrolling a few optimization steps with learnable step sizes and using the result as a training target—is potentially interesting, and the paper is unusually explicit in presenting its algorithm. However, the central experimental claims are not credible as reported, and the paper never defines the inference-time protocol for a module that requires ground-truth labels. If the method were re-evaluated with a clearly label-free inference protocol and realistic baselines, it might have value as a convergence-acceleration technique; in its current form, the evidence does not support the stated claims.

major comments (4)
  1. [Section IV-B and Algorithm 1] The inference-time protocol is never specified. O_POT in Eq. (12) is produced by K gradient-descent steps toward the ground-truth one-hot labels P_Y (Eqs. (4) and (11)), and O_POTGui in Eq. (13) is a weighted blend of the original logits with O_POT. Table II is titled "average inference performance," but Algorithm 1 describes only the training loop and does not state what the deployed model outputs at evaluation. If O_POT or O_POTGui is evaluated, ground-truth labels enter the forward computation directly and the 99.99 mIoU is trivial label leakage; if only the plain logits O from Eq. (2) are evaluated, no argument or ablation (e.g., sigma=1 vs. sigma=0.5 at eval) explains why optimizing Eq. (14) transfers to the unaugmented forward pass. The headline result is therefore not well-defined.
  2. [Table II and Section IV-B] The central experimental numbers are not credible. LVM+POTGui reaches 99.99 mIoU on Cityscapes, while the same table reports DeepLabv3+ at 69.04, SegFormer at 39.37, and BiSeNetV2 at 33.63. These baseline values are far below their published results on the same datasets, and 99.99 mIoU implies near-perfect per-pixel agreement across 19 classes, which no published segmentation method achieves. The paper must provide the full evaluation setup—input resolution, crop/stride, validation split, test-time augmentation, number of runs, and standard deviations—before any comparison can be interpreted.
  3. [Section III-B, Eqs. (7)-(10)] The derivation of the softmax cross-entropy gradient is incorrect as written. Eq. (8) omits the 1/N normalization, Eq. (9) keeps only the diagonal term of the softmax Jacobian, and Eq. (10) mixes the class index c on the left with the summation index t. The subsequent identity sum_t P_X,t P_Y,t = P_X,c only holds when P_Y is one-hot at class c, which is not articulated. Although Eq. (11) is the standard final result for one-hot labels, the presented steps do not rigorously establish it, so a central technical derivation needs correction.
  4. [Section III-C and Eq. (14)] The method is described as a "posterior optimization trajectory," but O_POT is not a predicted trajectory: it is constructed from the ground-truth labels P_Y at every unrolled step and is then used in a cross-entropy loss against the same P_Y. This makes the training signal partly answer-dependent, and the paper provides no theoretical analysis or ablation (e.g., detaching O_POT, using predicted pseudo-labels, or varying sigma during training/evaluation) to show whether the claimed speedup comes from this label-conditioned lookahead or from some other effect.
minor comments (6)
  1. [Abstract and Section I] There are typos such as "the the AD model" in the abstract and "P osterior" in Section I; the manuscript needs a careful proofreading pass.
  2. [Section IV-B] The heading "LVM+POTGui pk Benchmarks" contains a stray "pk" and should be corrected.
  3. [Table III] The class name "Signsymbol" should be checked against the standard CamVid class list; the paper also reports 0.00 IoU for several classes in baselines, which is unusual and should be explained (e.g., class never predicted due to evaluation protocol).
  4. [Figure 5] The ablation on the number of POTGen layers has no axis labels or numerical values, so claims such as "50 layers performs almost the same as 60 layers" are not quantitatively supported.
  5. [Section III-A and Related Work] Calling iGPT a Large Vision Model is unconventional, and the paper does not compare with more recent frozen backbones used for semantic segmentation; a brief positioning or comparison with such models would strengthen the claims.
  6. [Implementation Details] No code, checkpoints, random seeds, or per-run statistics are provided, which limits reproducibility of a paper whose main claims are experimental.

Circularity Check

1 steps flagged · score 6.0 of 10

The label-guided POTGen in Eqs. (4)/(11)/(12) makes O_POTGui a function of the ground-truth P_Y, and the paper never states that P_Y is withheld during inference; the reported 99.99 mIoU may therefore be label leakage rather than a model prediction.

  1. other [Section III-B, Eqs. (4), (11), (12); Section III-C, Eqs. (13), (14); Algorithm 1; Table II]
    "O^{(j)}_{POT}=O^{(j)}_{K}=O^{(j)}_{0}-\sum_{k=1}^{K}\eta_{k}\alpha_{k}\nabla_{O}L_{CE}(P^{(j)}_{Y},O^{(j)}_{k-1}) (Eq. 12); O^{(j)}_{POTGui}=\sigma O^{(j)}+(1-\sigma)O^{(j)}_{POT}, \sigma\in[0,1] (Eq. 13); 'we use the summation logits O^{(j)}_{POTGui} to take place of original logits O^{(j)} in Eq. (3) to calculate the loss to optimize the model by back propagation' (Eq. 14). Table II lists 'LVM+POTGui (Ours) 99.99 99.99 99.99 99.99 82.06 89.52 97.70 85.51' under 'Average inference performance comparison.'"

    The 'posterior optimization trajectory' is not predicted from the model; Eq. (11) gives each inner update as (P_{X}-P_{Y})/N, so O_{POT} in Eq. (12) is, by construction, obtained by moving logits toward the ground-truth one-hot P_{Y}. Eq. (13) then mixes this label-dependent quantity with the raw logits, and Eq. (14) minimizes cross-entropy against the same P_{Y}. The paper reports Table II as 'inference performance' but never states that POTGen or P_{Y} is removed at evaluation. If the label-dependent O_{POTGui} is used to produce the reported masks, the 99.99 mIoU is simply the answer key inserted into the forward computation, so the 'prediction' reduces to its input P_{Y}.

full rationale

Most of the paper's architecture—frozen iGPT backbone, ASSP head, standard cross-entropy baseline—is self-contained and benchmarked against external methods; the self-citations ([5]–[7], [30], [31]) are either related-work placements or an 'inspired by' pointer for unrolling optimization as layers, and none is load-bearing as an externally imported uniqueness theorem. The decisive issue is the label-dependent POTGen: Eqs. (4), (11), and (12) define O_POT by gradient descent toward P_Y, and Eqs. (13) and (14) make the optimized logits and the loss explicit functions of the same ground-truth labels. The headline Table II is captioned 'inference performance,' but the paper never specifies whether P_Y is withheld during evaluation; if it is not withheld, the near-perfect scores are label leakage by construction, and if it is withheld, the result is unexplained because no ablation separates sigma*O from (1-sigma)*O_POT. This conditional leakage strikes at the central claim, so the paper is partially circular/leakage-dependent, though the derivation is not a formal equivalence and the experimental framework is otherwise independent. Score 6 reflects one central 'prediction' that may reduce by construction, not a fully self-referential paper.

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

The central claim rests on a handful of fitted or hand-chosen quantities: the per-layer step sizes, the blend weight sigma, the number of unrolled layers K, and the choice of LVM feature layers. The method also assumes that label-guided logit refinement is a valid training target and that evaluation does not use labels at inference. These are not supported by external evidence.

free parameters (4)
  • alpha_k, eta_k per-layer step sizes = not reported
    Learnable parameters in Eq. (4) optimized by backpropagation; no final values are given, and they control the K-step logit update that defines O_POT.
  • sigma blend weight = 0.5
    Set by hand in Section IV-A.3; Eq. (13) mixes raw and refined logits, so the central loss depends on this choice.
  • K number of POTGen layers = not fixed; tested up to 60 in Fig. 5
    Performance and convergence speed vary with K; no principled selection criterion and no held-out validation are described.
  • feature-layer choice (LVM_Middle_4) = middle four layers of iGPT
    Selected after comparing six feature variants on the same benchmark datasets in Table I; this is a post-hoc model selection that conditions the main results.
assumptions (4)
  • standard math Cross-entropy gradients in logit space drive the logits toward the one-hot ground truth (Eq. 4, Eq. 11).
    Relies on the standard softmax cross-entropy gradient, though the derivation in Eqs. (7)-(10) mishandles off-diagonal softmax terms.
  • ad hoc to paper Unrolling K gradient steps on logits and blending with the original logits yields a valid training target for the perception head.
    This is the core premise of POTGui; no proof is given that backpropagating through this oracle-derived target improves generalization.
  • domain assumption The frozen LVM backbone features are sufficient for street scene understanding.
    The paper uses iGPT in zero-shot mode (Eq. 1) without fine-tuning; all learning is confined to the ASSP head.
  • domain assumption Evaluation is performed on held-out data without ground-truth labels in the loop at inference.
    This is never stated in Section IV; the reported near-perfect scores are only plausible if labels leak into the POTGen computation.
invented entities (1)
  • Posterior Optimization Trajectory (POT)
    purpose: An oracle-like sequence of future logit updates computed from ground-truth labels to guide current training (Eq. 12).
    No falsifiable handle outside the paper; it is constructed from the labels it later optimizes against.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Enhancing Large Vision Model in Street Scene Semantic Understanding through Leveraging Posterior Optimization Trajectory." pith.science (2026). https://pith.science/paper/BY4T3NKE

@misc{pith2026250101710,
  author       = {Pith},
  title        = {Pith review of: Enhancing Large Vision Model in Street Scene Semantic Understanding through Leveraging Posterior Optimization Trajectory},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/BY4T3NKE}},
  note         = {Machine review of arXiv:2501.01710}
}
read the original abstract

To improve the generalization of the autonomous driving (AD) perception model, vehicles need to update the model over time based on the continuously collected data. As time progresses, the amount of data fitted by the AD model expands, which helps to improve the AD model generalization substantially. However, such ever-expanding data is a double-edged sword for the AD model. Specifically, as the fitted data volume grows to exceed the the AD model's fitting capacities, the AD model is prone to under-fitting. To address this issue, we propose to use a pretrained Large Vision Models (LVMs) as backbone coupled with downstream perception head to understand AD semantic information. This design can not only surmount the aforementioned under-fitting problem due to LVMs' powerful fitting capabilities, but also enhance the perception generalization thanks to LVMs' vast and diverse training data. On the other hand, to mitigate vehicles' computational burden of training the perception head while running LVM backbone, we introduce a Posterior Optimization Trajectory (POT)-Guided optimization scheme (POTGui) to accelerate the convergence. Concretely, we propose a POT Generator (POTGen) to generate posterior (future) optimization direction in advance to guide the current optimization iteration, through which the model can generally converge within 10 epochs. Extensive experiments demonstrate that the proposed method improves the performance by over 66.48\% and converges faster over 6 times, compared to the existing state-of-the-art approach.

Figures

Figures reproduced from arXiv: 2501.01710 by the authors.

Figure 1
Figure 1. Illustration of the proposed POT-guided LVM-driven street scene semantic understanding method. [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Illustration of the proposed POTGui optimization [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 4
Figure 4. Illustration of performance of LVM+POTGui against existing state-of-the-art benchmarks on Cityscapes and CamVid. [PITH_FULL_IMAGE:figures/full_fig_p005_4.png] view at source ↗
Figures from the paper (2 more)
Figure 5
Figure 5. Figure 5: The effect of the number of layers on POTGui. [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 6
Figure 6. Figure 6: Real driving test of LVM+POTGui on Apolloscapes [PITH_FULL_IMAGE:figures/full_fig_p006_6.png]

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. Label Anything: An Interpretable, High-Fidelity and Prompt-Free Annotator

    cs.RO 2025-02 reject novelty 4.0 of 10

    The paper claims a single-seed, prompt-free annotator with 99.99% mIoU, but the described inference pipeline bakes the seed's ground-truth label into the output for every image.

Reference graph

Works this paper leans on

39 extracted references · 20 canonical work pages · cited by 1 Pith paper

  1. [1]

    Semi-supervised active learning for semantic segmentation in unknown environments using informative path planning,

    J. Rückin, F. Magistri, C. Stachniss, and M. Popovi ´c, “Semi-supervised active learning for semantic segmentation in unknown environments using informative path planning,” IEEE Robotics and Automation Letters, vol. 9, no. 3, pp. 2662–2669, 2024

    2024

  2. [2]

    Lightweight semantic segmentation network for semantic scene understanding on low-compute devices,

    H. Son and J. Weiland, “Lightweight semantic segmentation network for semantic scene understanding on low-compute devices,” in 2023 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2023, pp. 62–69

    2023

  3. [3]

    Ida: Informed domain adaptive semantic segmentation,

    Z. Chen, Z. Ding, J. M. Gregory, and L. Liu, “Ida: Informed domain adaptive semantic segmentation,” in 2023 IEEE/RSJ International Conference on Intelligent Robots and Systems(IROS) , 2023, pp. 90–97

    work page 2023

  4. [4]

    Featdanet: Feature-level domain adaptation network for semantic segmentation,

    J. Li, W. Shi, D. Zhu, G. Zhang, X. Zhang, and J. Li, “Featdanet: Feature-level domain adaptation network for semantic segmentation,” in 2023 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2023, pp. 3873–3880

    work page 2023

  5. [5]

    Fe- drc: A rapid-converged hierarchical federated learning framework in street scene semantic understanding,

    W.-B. Kou, Q. Lin, M. Tang, S. Wang, G. Zhu, and Y .-C. Wu, “Fe- drc: A rapid-converged hierarchical federated learning framework in street scene semantic understanding,” in 2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) . IEEE, 2024, pp. 2578–2585

    work page 2024

  6. [6]

    Fast-convergent and communication-alleviated heterogeneous hierarchical federated learning in autonomous driving,

    W.-B. Kou, Q. Lin, M. Tang, R. Ye, S. Wang, G. Zhu, and Y .- C. Wu, “Fast-convergent and communication-alleviated heterogeneous hierarchical federated learning in autonomous driving,” arXiv preprint arXiv:2409.19560, 2024

    arXiv 2024

  7. [7]

    An adverse weather-immune scheme with unfolded regularization and foundation model knowledge distillation for street scene understand- ing,

    W.-B. Kou, G. Zhu, R. Ye, S. Wang, Q. Lin, M. Tang, and Y .-C. Wu, “An adverse weather-immune scheme with unfolded regularization and foundation model knowledge distillation for street scene understand- ing,” arXiv preprint arXiv:2409.14737 , 2024

    arXiv 2024

  8. [8]

    Wedge: Web-image assisted domain generalization for semantic segmentation,

    N. Kim, T. Son, J. Pahk, C. Lan, W. Zeng, and S. Kwak, “Wedge: Web-image assisted domain generalization for semantic segmentation,” in 2023 IEEE International Conference on Robotics and Automation (ICRA), 2023, pp. 9281–9288

    work page 2023

Show all 39 references
  1. [9]

    Striking the right balance: Recall loss for semantic segmentation,

    J. Tian, N. C. Mithun, Z. Seymour, H.-P. Chiu, and Z. Kira, “Striking the right balance: Recall loss for semantic segmentation,” in 2022 International Conference on Robotics and Automation (ICRA) , 2022, pp. 5063–5069

    2022

  2. [10]

    Prompt, plan, perform: Llm-based humanoid control via quantized imitation learning,

    J. Sun, Q. Zhang, Y . Duan, X. Jiang, C. Cheng, and R. Xu, “Prompt, plan, perform: Llm-based humanoid control via quantized imitation learning,” in 2024 IEEE International Conference on Robotics and Automation (ICRA) , 2024, pp. 16 236–16 242

    2024

  3. [11]

    Prompting multi-modal tokens to enhance end-to-end autonomous driving imitation learning with llms,

    Y . Duan, Q. Zhang, and R. Xu, “Prompting multi-modal tokens to enhance end-to-end autonomous driving imitation learning with llms,” in 2024 IEEE International Conference on Robotics and Automation (ICRA), 2024, pp. 6798–6805

    2024

  4. [12]

    Minigpt-4: En- hancing vision-language understanding with advanced large language models,

    D. Zhu, J. Chen, X. Shen, X. Li, and M. Elhoseiny, “Minigpt-4: En- hancing vision-language understanding with advanced large language models,” CoRR, vol. abs/2304.10592, 2023

    2023 arXiv

  5. [13]

    Training language models to follow instructions with human feedback,

    L. Ouyang, J. Wu, X. Jiang, D. Almeida, C. L. Wainwright, P. Mishkin, C. Zhang, S. Agarwal, K. Slama, A. Ray, J. Schulman, J. Hilton, F. Kelton, L. Miller, M. Simens, A. Askell, P. Welinder, P. F. Chris- tiano, J. Leike, and R. Lowe, “Training language models to follow instruc...

    2022

  6. [14]

    Structgpt: A general framework for large language model to reason over structured data,

    J. Jiang, K. Zhou, Z. Dong, K. Ye, W. X. Zhao, and J.-R. Wen, “Structgpt: A general framework for large language model to reason over structured data,” 2023

    2023

  7. [15]

    Chain-of-thought prompting elicits reasoning in large language models,

    J. Wei, X. Wang, D. Schuurmans, M. Bosma, B. Ichter, F. Xia, E. H. Chi, Q. V . Le, and D. Zhou, “Chain-of-thought prompting elicits reasoning in large language models,” in NeurIPS, 2022

    2022

  8. [16]

    Visual language maps for robot navigation,

    C. Huang, O. Mees, A. Zeng, and W. Burgard, “Visual language maps for robot navigation,” in 2023 IEEE International Conference on Robotics and Automation (ICRA) . IEEE, 2023, pp. 10 608–10 615

    2023

  9. [17]

    Class semantics modulation for open-set instance segmentation,

    Y . Yang, Z. Zhou, J. Wu, Y . Wang, and R. Xiong, “Class semantics modulation for open-set instance segmentation,” IEEE Robotics and Automation Letters, vol. 9, no. 3, pp. 2240–2247, 2024

    2024

  10. [18]

    Unseen object few-shot semantic segmentation for robotic grasping,

    X. Liu, Y . Zhang, and D. Shan, “Unseen object few-shot semantic segmentation for robotic grasping,” IEEE Robotics and Automation Letters, vol. 8, no. 1, pp. 320–327, 2023

    2023

  11. [19]

    Cekd: Cross-modal edge-privileged knowledge distillation for semantic scene understanding using only thermal images,

    Z. Feng, Y . Guo, and Y . Sun, “Cekd: Cross-modal edge-privileged knowledge distillation for semantic scene understanding using only thermal images,” IEEE Robotics and Automation Letters , vol. 8, no. 4, pp. 2205–2212, 2023

    2023

  12. [20]

    Navya3dseg - navya 3d semantic segmentation dataset design & split generation for autonomous vehicles,

    A. Almin, L. Lemarié, A. Duong, and B. R. Kiran, “Navya3dseg - navya 3d semantic segmentation dataset design & split generation for autonomous vehicles,” IEEE Robotics and Automation Letters , vol. 8, no. 9, pp. 5584–5591, 2023

    2023

  13. [21]

    Fdlnet: Boosting real- time semantic segmentation by image-size convolution via frequency domain learning,

    Q. Yan, S. Li, C. Liu, M. Liu, and Q. Chen, “Fdlnet: Boosting real- time semantic segmentation by image-size convolution via frequency domain learning,” in 2023 IEEE International Conference on Robotics and Automation (ICRA) , 2023

    2023

  14. [22]

    Decoupling features in hierarchical propagation for video object segmentation,

    Z. Yang and Y . Yang, “Decoupling features in hierarchical propagation for video object segmentation,” in Advances in Neural Information Processing Systems (NeurIPS) , 2022

    2022

  15. [23]

    Rethinking semantic segmentation: A prototype view,

    T. Zhou, W. Wang, E. Konukoglu, and L. Van Gool, “Rethinking semantic segmentation: A prototype view,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2022, pp. 2582–2593

    2022

  16. [24]

    Segformer: Simple and efficient design for semantic segmen- tation with transformers,

    E. Xie, W. Wang, Z. Yu, A. Anandkumar, J. M. Alvarez, and P. Luo, “Segformer: Simple and efficient design for semantic segmen- tation with transformers,” in Neural Information Processing Systems (NeurIPS), 2021

    2021

  17. [25]

    Understanding bird’s-eye view of road semantics using an onboard camera,

    Y . B. Can, A. Liniger, O. Unal, D. Paudel, and L. Van Gool, “Understanding bird’s-eye view of road semantics using an onboard camera,” IEEE Robotics and Automation Letters , vol. 7, no. 2, pp. 3302–3309, 2022

    2022

  18. [26]

    Feddrive: Generalizing federated learning to semantic segmentation in autonomous driving,

    L. Fantauzzo, E. Fanì, D. Caldarola, A. Tavera, F. Cermelli, M. Ci- ccone, and B. Caputo, “Feddrive: Generalizing federated learning to semantic segmentation in autonomous driving,” in Proceedings of the 2022 IEEE/RSJ International Conference on Intelligent Robots and Systems, 2022

    2022

  19. [27]

    Communication resources constrained hierarchical feder- ated learning for end-to-end autonomous driving,

    W.-B. Kou, S. Wang, G. Zhu, B. Luo, Y . Chen, D. W. Kwan Ng, and Y .-C. Wu, “Communication resources constrained hierarchical feder- ated learning for end-to-end autonomous driving,” in 2023 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) , 2023, pp....

    2023

  20. [28]

    Generative pretraining from pixels,

    M. Chen, A. Radford, R. Child, J. Wu, H. Jun, D. Luan, and I. Sutskever, “Generative pretraining from pixels,” in International conference on machine learning . PMLR, 2020, pp. 1691–1703

    2020

  21. [29]

    Deeplab: Semantic image segmentation with deep convolutional nets, atrous convolution, and fully connected crfs,

    L.-C. Chen, G. Papandreou, I. Kokkinos, K. Murphy, and A. L. Yuille, “Deeplab: Semantic image segmentation with deep convolutional nets, atrous convolution, and fully connected crfs,” IEEE transactions on pattern analysis and machine intelligence , vol. 40, no. 4, pp. 834– 848, 2017

    2017

  22. [30]

    Communication-efficient activity detection for cell-free massive mimo: An augmented model-driven end-to-end learning framework,

    Q. Lin, Y . Li, W.-B. Kou, T.-H. Chang, and Y .-C. Wu, “Communication-efficient activity detection for cell-free massive mimo: An augmented model-driven end-to-end learning framework,” IEEE Transactions on Wireless Communications , pp. 1–1, 2024

    2024

  23. [31]

    Communication- efficient joint signal compression and activity detection in cell-free massive mimo,

    Q. Lin, Y . Li, W.-B. Kou, T. Chang, and Y .-C. Wu, “Communication- efficient joint signal compression and activity detection in cell-free massive mimo,” in ICC 2023-IEEE International Conference on Communications. IEEE, 2023, pp. 5030–5035

    2023

  24. [32]

    The cityscapes dataset for semantic urban scene understanding,

    M. Cordts, M. Omran, S. Ramos, T. Rehfeld, M. Enzweiler, R. Benen- son, U. Franke, S. Roth, and B. Schiele, “The cityscapes dataset for semantic urban scene understanding,” in Proc. of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) , 2016

    2016

  25. [33]

    Segmentation and recognition using structure from motion point clouds,

    G. J. Brostow, J. Shotton, J. Fauqueur, and R. Cipolla, “Segmentation and recognition using structure from motion point clouds,” in Com- puter Vision–ECCV 2008: 10th European Conference on Computer Vision, Marseille, France, October 12-18, 2008, Proceedings, Part I

    2008

  26. [34]

    Springer, 2008, pp. 44–57

    2008

  27. [35]

    The apolloscape open dataset for autonomous driving and its application,

    P. Wang, X. Huang, X. Cheng, D. Zhou, Q. Geng, and R. Yang, “The apolloscape open dataset for autonomous driving and its application,” IEEE transactions on pattern analysis and machine intelligence , 2019

    2019

  28. [36]

    Carla: An open urban driving simulator,

    A. Dosovitskiy, G. Ros, F. Codevilla, A. Lopez, and V . Koltun, “Carla: An open urban driving simulator,” in Proceedings of The 1st Annual Conference on Robot Learning , Oct. 2017, pp. 1–16

    2017

  29. [37]

    Bisenet v2: Bilateral network with guided aggregation for real-time semantic segmentation,

    C. Yu, C. Gao, J. Wang, G. Yu, C. Shen, and N. Sang, “Bisenet v2: Bilateral network with guided aggregation for real-time semantic segmentation,” International Journal of Computer Vision , vol. 129, pp. 3051–3068, 2021

    2021

  30. [38]

    Segnet: A deep convolutional encoder-decoder architecture for image segmentation,

    V . Badrinarayanan, A. Kendall, and R. Cipolla, “Segnet: A deep convolutional encoder-decoder architecture for image segmentation,” IEEE transactions on pattern analysis and machine intelligence , vol. 39, no. 12, pp. 2481–2495, 2017

    2017

  31. [39]

    Encoder-decoder with atrous separable convolution for semantic image segmentation,

    L.-C. Chen, Y . Zhu, G. Papandreou, F. Schroff, and H. Adam, “Encoder-decoder with atrous separable convolution for semantic image segmentation,” 2018

    2018

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.