Pith. sign in

REVIEW 3 major objections 6 minor 82 references

Sketch Down the FLOPs: Towards Efficient Networks for Human Sketch

T0 review · 3 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read The paper claims that efficient networks for human sketch are achievable by combining cross-modal knowledge distillation with a reinforcement-learned canvas-size selector, cutting the FLOPs of a VGG-16 FG-SBIR model from 40.18G to 0.254G…

desk verdict A solid first efficiency paper for FG-SBIR whose headline 'retaining accuracy' rests on point estimates without variance; the method is sound and the paper deserves peer review. read the letter →

arxiv 2505.23763 v1 pith:7PR7SCUJ submitted 2025-05-29 cs.CV

Aneeshan Sain , Subhajit Maity , Pinaki Nath Chowdhury , Subhadeep Koley , Ayan Kumar Bhunia , Yi-Zhe Song This is my paper
classification cs.CV
keywords efficientinferencehumansketchfine-grainedsketch-basedimageretrievalknowledgedistillationreinforcementlearningcanvas-sizeselectionFLOPsreductionMobileNetV2
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

Human sketches are sparse, vector inputs, yet no efficient-inference method has been designed for them, and photo-efficient backbones such as MobileNetV2 lose about 37% relative accuracy when used off-the-shelf for fine-grained sketch-based image retrieval (FG-SBIR). The paper claims this gap can be closed by two plug-n-play components: a cross-modal knowledge distillation that transfers the pairwise distance structure of a large teacher (VGG-16) into a MobileNetV2 student, and a reinforcement-learned canvas-size selector that renders each vector sketch at the smallest resolution that preserves retrieval. Together they reduce FLOPs from 40.18G to 0.254G (99.37%) on the ShoeV2 benchmark while keeping Top-1 accuracy essentially unchanged (33.03% vs 32.77%). If correct, this makes FG-SBIR, a task with direct commercial value, deployable on edge devices for the first time.

What carries the argument

Two components carry the argument. SketchyNetV1 is a relational knowledge-distillation scheme: a MobileNetV2 student is trained with a triplet loss plus a Huber loss on the pairwise $\ell^2$ distances ($d_{sp}$, $d_{sn}$, $d_{pn}$) between sketch, positive-photo, and negative-photo embeddings of a frozen VGG-16 teacher, so the student inherits the teacher's embedding geometry rather than a fragile feature regression. SketchyNetV2 adds a canvas-size selector: a one-layer GRU that reads the vector sketch (sequences of $x,y$ and pen-state coordinates, capped by Douglas-Peucker simplification) and samples a canvas size from $\{32,64,128,256\}$; the retrieval network acts as a critic, and policy-gradient training maximizes a reward that combines inverse rank and negative triplet loss with a normalized FLOP penalty. The selector is what exploits sketch abstraction, since sketches rendered at $32\times32$ retain far more retrieval accuracy than photos do at the same resolution.

What would settle it

A reader could render every test sketch at all four canvas sizes, run the full teacher network on each, and record the size giving the best retrieval rank; if the learned selector agrees with that per-sketch argmax far less often than the reported cross-model agreement rate, or if its choices correlate with student-specific errors, the claim that it selects the content-optimal canvas is refuted.

Watch

Extended reading notes

Core claim

The paper's central discovery is that the two properties that make sketches hard for photo models—their fine-grained detail and their abstraction—are exactly the properties an efficient sketch network should exploit. A small student trained to reproduce the teacher's inter-feature distances for sketch-photo-negative triplets, rather than its raw features, can match a VGG-16 teacher's retrieval accuracy at a fraction of the FLOPs; and because sketches are vector graphics that can be rendered at any size, a policy trained to pick the smallest canvas that keeps the retrieved rank high cuts the remaining compute by another factor of 3.28. The paper verifies the core numbers on four FG-SBIR datasets and shows the canvas selector transfers across teachers and datasets.

Load-bearing premise

The load-bearing premise is that the student's own retrieval score is a trustworthy proxy for the true minimal canvas size of a sketch; the paper reports cross-model agreement of 89.21% on selected sizes, but if the student systematically favors small canvases that hide hard negatives, the FLOPs savings would be tuned to the student's blind spots rather than to sketch content.

Editorial extensions

If this is right

  • Off-the-shelf efficient photo models are not a shortcut for sketch retrieval: the paper's pilot shows MobileNetV2 at 0.833G reaches only 20.85% Top-1 on ShoeV2, versus 33.03% for VGG-16, and fixed low resolution does not close that gap.
  • The two components compose: SketchyNetV1 alone reduces FLOPs by roughly 48x with a marginal accuracy drop, and the canvas selector then cuts another factor of 3.28 while keeping accuracy at 32.77% versus 33.03%.
  • Accuracy retention holds across different teachers and datasets: SketchyNetV2 variants of existing FG-SBIR models stay within about 1% of their teachers while using far fewer FLOPs, and a canvas selector trained on one dataset and teacher transfers to another with a small loss (e.g., 53.14% vs 55.32% on ChairV2).
  • The canvas selector is not tied to FG-SBIR: plugged into a standard sketch-recognition pipeline on QuickDraw it drops FLOPs from 5.28G to 0.261G while keeping accuracy at 68.14% versus 68.71%.
  • Because the selector is a separate module, accuracy-focused and transformer-based retrieval models can be fitted with it; the supplementary material shows transformer teachers maintain near-teacher accuracy in their SketchyNetV2 variants.

Reading between the lines

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

  • Editorial extension: the selector operates entirely in vector space, so the same abstraction-aware resolution scheduling should transfer to other vector line inputs such as handwriting, technical drawings, and floor plans, where rendering cost and retrieval or recognition accuracy trade off similarly.
  • Editorial extension: the paper's future-work suggestion could be turned into a concrete test—use the per-sketch chosen canvas to label how much visual information a sketch truly needs, then train a photo resizer to shrink photos to that budget, extending the FLOPs savings to the photo gallery side of retrieval.
  • Editorial extension: because the reward uses the student's own rank and triplet loss, the selector may inherit the student's blind spots; a useful control experiment would be training the selector with rewards from the full teacher network and comparing the selected canvas sizes.
  • Editorial extension: the method composes with quantization and pruning—the paper reports 0.249G with pruning and about 9.1ms with INT8 on an iPhone 13—so the reported 160x reduction is not an upper bound, and an INT8-quantized SketchyNetV2 is a plausible sub-0.1G target.
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

3 major / 6 minor

Summary. The paper proposes two components for efficient fine-grained sketch-based image retrieval (FG-SBIR): SketchyNetV1, a MobileNetV2 student trained from a VGG-16 teacher via relational knowledge distillation on pairwise embedding distances, and SketchyNetV2, which adds a reinforcement-learning-based canvas-size selector that rasterizes each vector sketch at a predicted resolution to reduce FLOPs further. The headline result is a reduction from 40.18G to 0.254G FLOPs (99.37%) with reported Top-1 accuracy of 33.03% vs. 32.77% on QMUL-ShoeV2, plus analogous results on ChairV2, Sketchy, and FS-COCO. The paper includes pilot studies motivating the backbone choice and dynamic canvas-sizing, ablations of reward components and backbones, and cross-model and cross-dataset generalization tests of the canvas selector.

Significance. If the claims hold, this is a useful and timely contribution: it is, to my knowledge, the first systematic study of efficient inference specifically for FG-SBIR, and the two modules are generic enough to be attached to different teacher-student pairs. The paper's strengths are its breadth of evaluation (four FG-SBIR datasets plus categorical SBIR and sketch recognition), the explicit pilot studies that motivate the design, and the cross-model agreement experiments (89.21%, 94.21%, 95.64%) that partially discharge concerns about the RL selector fitting a single critic. The FLOPs reductions are consistent across all tested teachers, and the ablations in Table 3 justify the main design choices. However, the central claim of 'retaining accuracy' needs stronger statistical support before it can be taken as established.

major comments (3)
  1. [Table 2 and abstract] The headline claim of 'retaining the accuracy (33.03% vs 32.77%)' rests on single point estimates on 679 evaluation sketches in QMUL-ShoeV2; the 0.26-point difference is less than two correctly retrieved queries. No standard deviations, multiple seeds, or significance tests are reported anywhere in the paper. At n=679, the standard error of this accuracy difference is on the order of 2-3 percentage points, so the observed parity is compatible with a nontrivial true accuracy drop. The same issue affects all datasets in Table 2. I recommend reporting mean±std across at least three seeds, or releasing per-query outputs so that paired bootstrap confidence intervals can be computed; without this, the central 'retaining accuracy' claim is not statistically established.
  2. [Table 2, Sketchy and FS-COCO rows] Even ignoring statistical noise, the point estimates on two of the four datasets show non-negligible relative drops: Top-1 falls from 18.61% to 16.55% on Sketchy (-2.06 points, about 11% relative) and from 5.16% to 4.69% on FS-COCO (-0.47 points, about 9% relative). The abstract and conclusion describe the result as 'retaining the accuracy,' which is accurate only for ShoeV2 (and arguably ChairV2). The paper should either present these as an explicit accuracy-efficiency trade-off or add a qualifier such as 'within a small margin on most datasets.' This is load-bearing because the contribution is framed as accuracy-neutral compression.
  3. [Section 4.2, Eqs. (4)-(6)] The RL reward for the canvas-size selector is computed from the student's own rank and triplet loss, and the final evaluation of SketchyNetV2 uses the same student. This creates a risk that the selector is tuned to the student's resolution-dependent blind spots rather than to the sketch's intrinsic abstraction level. The cross-model agreement tests in Section 5.3 substantially mitigate this risk, and I credit them; however, agreement on discrete canvas sizes does not fully guarantee that the chosen sizes preserve accuracy on an independent critic. I ask for one additional experiment targeting the headline model: train the canvas-size selector for the B-VGG16-SN-based SketchyNetV2 setup using the teacher FT as the critic (instead of the student), then report the student's Top-1 and FLOPs. The existing teacher-as-critic experiment in Section 5.3 is for the Triplet-SN setting, not for the B-VGG16-SN model whose accuracy retention is the headline claim.
minor comments (6)
  1. [Abstract vs. Conclusion] The abstract reports a FLOPs reduction of '97.96% percent' and the conclusion reports '97.92 (84.9)%'; the two numbers should be reconciled. Also, '97.96% percent' contains a redundant 'percent.'
  2. [Section 5.3] The sentence comparing offset-coordinate and absolute vector formats says 'Turns out the former is better for encoding,' but Table 3 shows the absolute-coordinate version (Ours) achieving 32.77% Top-1 versus 32.61% for offset-coordinate; the text should say 'the latter' or correct the table.
  3. [Table 1] The caption does not state the input resolution at which FLOPs, parameters, and per-sample time are measured; given the paper's focus on 256×256 as the full resolution, the table should specify this explicitly.
  4. [Figure 2] The right panel's bar heights are described in the text as percentages of sketches achieving perfect retrieval at each canvas size, normalized by those achieving perfect retrieval at 256×256, but the axis label '% sketches' is ambiguous; the caption should explain this normalization.
  5. [Section 5.3, Comparing Inference Times] The reported full-sketch inference times for the SoTAs (37, 38, 36, 42 ms) are not directly comparable to the per-sample feature extraction times in Table 1 (e.g., VGG-16 at 84 ms), since the SoTAs use different backbones and possibly different hardware; please clarify what is being measured and on which hardware.
  6. [General] No code or pre-trained models are released. Given the number of hyperparameters (λ, λF, λr, λTri, β, Tmax) and the RL training procedure, releasing code would substantially strengthen reproducibility and would also allow readers to compute bootstrap confidence intervals for the headline accuracy claim.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the KD and RL components are trained on held-out data and benchmarked against external baselines.

full rationale

The derivation chain is self-contained. SketchyNetV1's distillation objective (Eq. 3) is a standard combination of triplet loss and relational KD (Eq. 2), where the teacher is an externally trained VGG-16 FG-SBIR model and the student's accuracy is measured on held-out test splits. The RL canvas selector (Sec. 4.2) frames canvas-size selection as a policy optimization problem using the fixed student as a critic; the reward (Eqs. 4-6) trades retrieval rank against FLOPs and is optimized via policy gradient (Eq. 7). This is a training objective, not a derivation that presupposes the reported accuracy retention. The retained-accuracy claim (33.03 vs 32.77 on ShoeV2) is a measured test-set result benchmarked against external SoTAs and pruning/quantization baselines in Table 2. The paper directly probes the potential arbitrariness of its canvas selector through cross-model agreement experiments (89.21%, 94.21%, 95.64% agreement across different critics), showing the chosen sizes are not idiosyncratic to one model. The remaining weakness - absence of variance or significance estimates for small accuracy differences - is a statistical robustness concern, not a circularity. No load-bearing self-citation or imported uniqueness theorem appears; self-citations are contextual and do not substitute for the paper's own experiments.

Assumptions & free parameters 8 free parameters · 6 assumptions · 0 invented entities

The framework relies on several domain assumptions about sketch sparsity and vector rasterization, plus hand-set hyperparameters; no new physical entities are introduced.

free parameters (8)
  • lambda (KD weight) = 0.5
    Chosen by grid in Sec 5.2 (Fig 5) and ablation; balances triplet loss and relational KD loss in Eq. 3.
  • lambda_F (FLOPs reward weight) = 0.35
    Chosen by sweeping lambda_F in Fig 5 left; controls trade-off between accuracy reward and compute reward in Eq. 6.
  • lambda_r (rank reward weight) = 0.4
    Hyper-parameter in Eq. 4, empirically set in Implementation Details.
  • lambda_Tri (triplet loss reward weight) = 0.48
    Hyper-parameter in Eq. 4, empirically set in Implementation Details.
  • Triplet margin m = 0.2
    Margin in Eq. 1 set in Implementation Details.
  • Huber beta = 1
    Smoothing parameter in Eq. 2, set in Implementation Details.
  • Canvas selector embedding size = 128
    GRU hidden state dimension in Sec. 4.2, chosen empirically.
  • Tmax for vector sketches = 100
    Maximum sequence length after Douglas-Peucker simplification, set in Implementation Details.
assumptions (6)
  • domain assumption Triplet ranking loss with margin m is an adequate objective for FG-SBIR.
    Used as the base supervision for teacher, student, and RL critic (Eq. 1); standard in the cited FG-SBIR literature.
  • domain assumption A sketch rendered at lower resolution retains enough semantic information for retrieval.
    Pilot study (Sec. 3, Fig. 2) supports this, but the canvas selector optimizes on top of it.
  • domain assumption Vector sketches can be rasterized to any canvas size at negligible cost and without altering content.
    Assumed throughout Sec. 4.2; rasterization Rc(·) is treated as free and exact.
  • domain assumption Douglas-Peucker simplification to Tmax=100 preserves the visual information needed for canvas selection.
    Sec. 4.2 limits sequence length; no quantitative validation of information preservation is given.
  • domain assumption A single-layer GRU over the vector sequence can encode abstraction level sufficient for canvas-size prediction.
    Architecture choice in Sec. 4.2; ablation covers LSTM and Transformer but not the underlying representational assumption.
  • domain assumption The student's rank and triplet loss are valid reward signals for canvas selection.
    Eq. 4 uses 1/r and triplet loss; this is the source of the mild circularity because the same student is later evaluated.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Sketch Down the FLOPs: Towards Efficient Networks for Human Sketch." pith.science (2026). https://pith.science/paper/7PR7SCUJ

@misc{pith2026250523763,
  author       = {Pith},
  title        = {Pith review of: Sketch Down the FLOPs: Towards Efficient Networks for Human Sketch},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/7PR7SCUJ}},
  note         = {Machine review of arXiv:2505.23763}
}
read the original abstract

As sketch research has collectively matured over time, its adaptation for at-mass commercialisation emerges on the immediate horizon. Despite an already mature research endeavour for photos, there is no research on the efficient inference specifically designed for sketch data. In this paper, we first demonstrate existing state-of-the-art efficient light-weight models designed for photos do not work on sketches. We then propose two sketch-specific components which work in a plug-n-play manner on any photo efficient network to adapt them to work on sketch data. We specifically chose fine-grained sketch-based image retrieval (FG-SBIR) as a demonstrator as the most recognised sketch problem with immediate commercial value. Technically speaking, we first propose a cross-modal knowledge distillation network to transfer existing photo efficient networks to be compatible with sketch, which brings down number of FLOPs and model parameters by 97.96% percent and 84.89% respectively. We then exploit the abstract trait of sketch to introduce a RL-based canvas selector that dynamically adjusts to the abstraction level which further cuts down number of FLOPs by two thirds. The end result is an overall reduction of 99.37% of FLOPs (from 40.18G to 0.254G) when compared with a full network, while retaining the accuracy (33.03% vs 32.77%) -- finally making an efficient network for the sparse sketch data that exhibit even fewer FLOPs than the best photo counterpart.

Figures

Figures reproduced from arXiv: 2505.23763 by the authors.

Figure 1
Figure 1. Our SketchyNetV1 compresses existing heavy FG￾SBIR networks to deliver smaller models. Further enhanced via a canvas-selector module our SketchyNetV2 model minimises sketch-resolution dynamically to reduce FLOPs. sequential [23], abstract [63], or stroke-wise [10], style￾diversity [54], and data-scarcity [6, 21, 46]. This is partic￾ularly disappointing for the problem of fine-grained sketch￾based image retrieval (FG… view at source ↗
Figure 2
Figure 2. Pilot study for varying canvas-size (see text above). [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Two stage framework. (Left) SketchyNetV1: A smaller [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (1 more)
Figure 5
Figure 5. Figure 5: Impact of varying λF on Acc@1 and Acc@5 Architectural insights: To judge the efficiency of our student’s backbone, we explore a few cheaper backbone-networks (Types IV, V in [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

82 extracted references · 69 canonical work pages

  1. [56]

    Exploiting Unlabelled Photos for Stronger Fine-Grained SBIR

    Aneeshan Sain, Ayan Kumar Bhunia, Subhadeep Koley, Pinaki Nath Chowdhury, Soumitri Chattopadhyay, Tao Xi- ang, and Yi-Zhe Song. Exploiting Unlabelled Photos for Stronger Fine-Grained SBIR. In CVPR, 2023. 1, 12

    work page 2023

  2. [1]

    Decore: Deep compression with reinforcement learning

    Manoj Alwani, Yang Wang, and Vashisht Madhavan. Decore: Deep compression with reinforcement learning. In CVPR, 2022. 2

    2022

  3. [2]

    CLIP for All Things Zero- Shot Sketch-Based Image Retrieval, Fine-Grained or Not

    Aneeshan Sain and Ayan Kumar Bhunia and Pinaki Nath Chowdhury and Aneeshan Sain and Subhadeep Koley and Tao Xiang and Yi-Zhe Song. CLIP for All Things Zero- Shot Sketch-Based Image Retrieval, Fine-Grained or Not. In CVPR, 2023. 1

    2023

  4. [3]

    Do deep nets really need to be deep? NeurIPS, 2014

    Jimmy Ba and Rich Caruana. Do deep nets really need to be deep? NeurIPS, 2014. 2

    work page 2014

  5. [4]

    Label refinery: Improving ima- genet classification through label progression.arXiv preprint arXiv:1805.02641, 2018

    Hessam Bagherinezhad, Maxwell Horton, Mohammad Rastegari, and Ali Farhadi. Label refinery: Improving ima- genet classification through label progression.arXiv preprint arXiv:1805.02641, 2018. 2

    arXiv 2018

  6. [5]

    Pixelor: A competitive sketching ai agent

    Ayan Kumar Bhunia, Ayan Das, Umar Riaz Muhammad, Yongxin Yang, Timothy M Hospedales, Tao Xiang, Yulia Gryaditskaya, and Yi-Zhe Song. Pixelor: A competitive sketching ai agent. so you think you can sketch? ACM TOG,

    work page

  7. [6]

    Sketch less for more: On-the-fly fine-grained sketch based image retrieval

    Ayan Kumar Bhunia, Yongxin Yang, Timothy M Hospedales, Tao Xiang, and Yi-Zhe Song. Sketch less for more: On-the-fly fine-grained sketch based image retrieval. In CVPR, 2020. 1, 2, 3, 6

    work page 2020

  8. [7]

    More photos are all you need: Semi-supervised learning for fine- grained sketch based image retrieval

    Ayan Kumar Bhunia, Pinaki Nath Chowdhury, Aneeshan Sain, Yongxin Yang, Tao Xiang, and Yi-Zhe Song. More photos are all you need: Semi-supervised learning for fine- grained sketch based image retrieval. In CVPR, 2021. 1, 2, 3, 4

    work page 2021

Show all 82 references
  1. [8]

    Vectorization and rasterization: Self-supervised learning for sketch and handwriting

    Ayan Kumar Bhunia, Pinaki Nath Chowdhury, Yongxin Yang, Timothy M Hospedales, Tao Xiang, and Yi-Zhe Song. Vectorization and rasterization: Self-supervised learning for sketch and handwriting. In CVPR, 2021. 3

    2021

  2. [9]

    Sketching without worrying: Noise-tolerant sketch-based image retrieval

    Ayan Kumar Bhunia, Subhadeep Koley, Abdullah Faiz Ur Rahman Khilji, Aneeshan Sain, Pinaki Nath Chowdhury, Tao Xiang, and Yi-Zhe Song. Sketching without worrying: Noise-tolerant sketch-based image retrieval. In CVPR, 2022. 1, 2, 3, 7

    2022

  3. [10]

    Sketch2Saliency: Learning to Detect Salient Ob- jects from Human Drawings

    Ayan Kumar Bhunia, Subhadeep Koley, Amandeep Kumar, Aneeshan Sain, Pinaki Nath Chowdhury, Tao Xiang, and Yi- Zhe Song. Sketch2Saliency: Learning to Detect Salient Ob- jects from Human Drawings. In CVPR, 2023. 1

    2023

  4. [11]

    A simple framework for contrastive learning of visual representations

    Ting Chen, Simon Kornblith, Mohammad Norouzi, and Ge- offrey Hinton. A simple framework for contrastive learning of visual representations. In ICML, 2020. 3

    2020

  5. [12]

    Mobile- former: Bridging mobilenet and transformer

    Yinpeng Chen, Xiyang Dai, Dongdong Chen, Mengchen Liu, Xiaoyi Dong, Lu Yuan, and Zicheng Liu. Mobile- former: Bridging mobilenet and transformer. In CVPR,

  6. [13]

    Dynamic low-resolution distillation for cost- efficient end-to-end text spotting

    Ying Chen, Liang Qiao, Zhanzhan Cheng, Shiliang Pu, Yi Niu, and Xi Li. Dynamic low-resolution distillation for cost- efficient end-to-end text spotting. In ECCV, 2022. 5, 6, 7

    2022

  7. [14]

    Partially does it: Towards scene-level fg-sbir with partial input

    Pinaki Nath Chowdhury, Ayan Kumar Bhunia, Viswanatha Reddy Gajjala, Aneeshan Sain, Tao Xiang, and Yi-Zhe Song. Partially does it: Towards scene-level fg-sbir with partial input. In CVPR, 2022. 1, 2

    2022

  8. [15]

    Partially does it: Towards scene-level fg-sbir with partial input

    Pinaki Nath Chowdhury, Ayan Kumar Bhunia, Viswanatha Reddy Gajjala, Aneeshan Sain, Tao Xiang, and Yi-Zhe Song. Partially does it: Towards scene-level fg-sbir with partial input. In The IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2022. 6

    2022

  9. [16]

    Fs- coco: Towards understanding of freehand sketches of com- mon objects in context

    Pinaki Nath Chowdhury, Aneeshan Sain, Ayan Kumar Bhu- nia, Tao Xiang, Yulia Gryaditskaya, and Yi-Zhe Song. Fs- coco: Towards understanding of freehand sketches of com- mon objects in context. In ECCV, 2022. 5, 6

    2022

  10. [17]

    Democratising 2D Sketch to 3D Shape Retrieval Through Pivoting

    Chowdhury, Pinaki Nath and Bhunia, Ayan Kumar and Sain, Aneeshan and Koley, Subhadeep and Xiang, Tao and Song, Yi-Zhe. Democratising 2D Sketch to 3D Shape Retrieval Through Pivoting. In ICCV, 2023. 2

    2023

  11. [18]

    Livesketch: Query perturbations for guided sketch-based visual search

    John Collomosse, Tu Bui, and Hailin Jin. Livesketch: Query perturbations for guided sketch-based visual search. In CVPR, 2019. 2, 5

    2019

  12. [19]

    Binaryconnect: Training deep neural networks with binary weights during propagations

    Matthieu Courbariaux, Yoshua Bengio, and Jean-Pierre David. Binaryconnect: Training deep neural networks with binary weights during propagations. NeurIPS, 28, 2015. 2, 6

    2015

  13. [20]

    Doodle to search: Practical zero-shot sketch- based image retrieval

    Sounak Dey, Pau Riba, Anjan Dutta, Josep Llados, and Yi- Zhe Song. Doodle to search: Practical zero-shot sketch- based image retrieval. In CVPR, 2019. 2, 8

    2019

  14. [21]

    Semantically tied paired cycle consistency for zero-shot sketch-based image retrieval

    Anjan Dutta and Zeynep Akata. Semantically tied paired cycle consistency for zero-shot sketch-based image retrieval. In CVPR, 2019. 1

    2019

  15. [22]

    Deep learning with limited numerical precision

    Suyog Gupta, Ankur Agrawal, Kailash Gopalakrishnan, and Pritish Narayanan. Deep learning with limited numerical precision. In ICML, 2015. 2, 6, 8

    2015

  16. [23]

    A neural representation of sketch drawings

    David Ha and Douglas Eck. A neural representation of sketch drawings. In ICLR, 2018. 1, 5, 7, 8

    2018

  17. [24]

    Deep com- pression: Compressing deep neural networks with pruning, trained quantization and huffman coding

    Song Han, Huizi Mao, and William J Dally. Deep com- pression: Compressing deep neural networks with pruning, trained quantization and huffman coding. In ICLR, 2016. 2

    2016

  18. [25]

    Deep residual learning for image recognition

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition. In CVPR,

  19. [26]

    Distilling the knowledge in a neural network

    Geoffrey Hinton, Oriol Vinyals, and Jeff Dean. Distilling the knowledge in a neural network. In NeurIPS Deep Learning Workshop, 2014. 2

    2014

  20. [27]

    Densely connected convolutional net- works

    Gao Huang, Zhuang Liu, Laurens Van Der Maaten, and Kil- ian Q Weinberger. Densely connected convolutional net- works. In CVPR, 2017. 3

    2017

  21. [28]

    Robust pairwise learning with huber loss

    Shouyou Huang and Qiang Wu. Robust pairwise learning with huber loss. Journal of Complexity, 2021. 4, 12

    2021

  22. [29]

    Like what you like: Knowl- edge distill via neuron selectivity transfer

    Zehao Huang and Naiyan Wang. Like what you like: Knowl- edge distill via neuron selectivity transfer. arXiv preprint arXiv:1707.01219, 2017. 2

    2017 arXiv

  23. [30]

    Reinforcement learning: A survey

    Leslie Pack Kaelbling, Michael L Littman, and Andrew W Moore. Reinforcement learning: A survey. JAIR, 1996. 3, 5

    1996

  24. [31]

    Very deep convolu- tional networks for large-scale image recognition

    Andrew Zisserman Karen Simonyan. Very deep convolu- tional networks for large-scale image recognition. In ICLR,

  25. [32]

    How to handle sketch-abstraction in sketch-based image retrieval? In CVPR, 2024

    Subhadeep Koley, Ayan Kumar Bhunia, Aneeshan Sain, Pinaki Nath Chowdhury, Tao Xiang, and Yi-Zhe Song. How to handle sketch-abstraction in sketch-based image retrieval? In CVPR, 2024. 12 9

    2024

  26. [33]

    A2- rl: Aesthetics aware reinforcement learning for image crop- ping

    Debang Li, Huikai Wu, Junge Zhang, and Kaiqi Huang. A2- rl: Aesthetics aware reinforcement learning for image crop- ping. In CVPR, 2018. 3

    2018

  27. [34]

    Learning to learn cropping models for different aspect ratio require- ments

    Debang Li, Junge Zhang, and Kaiqi Huang. Learning to learn cropping models for different aspect ratio require- ments. In CVPR, 2020. 6

    2020

  28. [35]

    Deep variation- structured reinforcement learning for visual relationship and attribute detection

    Xiaodan Liang, Lisa Lee, and Eric P Xing. Deep variation- structured reinforcement learning for visual relationship and attribute detection. In CVPR, 2017. 3

    2017

  29. [36]

    Fixed point quantization of deep convolutional networks

    Darryl Lin, Sachin Talathi, and Sreekanth Annapureddy. Fixed point quantization of deep convolutional networks. In ICML, 2016. 2

    2016

  30. [37]

    Sketch-bert: Learning sketch bidirectional encoder rep- resentation from transformers by self-supervised learning of sketch gestalt

    Hangyu Lin, Yanwei Fu, Yu-Gang Jiang, and Xiangyang Xue. Sketch-bert: Learning sketch bidirectional encoder rep- resentation from transformers by self-supervised learning of sketch gestalt. In CVPR, 2020. 5

    2020

  31. [38]

    Deep sketch hashing: Fast free-hand sketch-based im- age retrieval

    Li Liu, Fumin Shen, Yuming Shen, Xianglong Liu, and Ling Shao. Deep sketch hashing: Fast free-hand sketch-based im- age retrieval. In CVPR, 2017. 2, 8

    2017

  32. [39]

    Efficientvit: Memory efficient vision transformer with cascaded group attention

    Xinyu Liu, Houwen Peng, Ningxin Zheng, Yuqing Yang, Han Hu, and Yixuan Yuan. Efficientvit: Memory efficient vision transformer with cascaded group attention. In CVPR,

  33. [40]

    Thinet: A filter level pruning method for deep neural network compression

    Jian-Hao Luo, Jianxin Wu, and Weiyao Lin. Thinet: A filter level pruning method for deep neural network compression. In ICCV, 2017. 2, 6

    2017

  34. [41]

    Pruning convolutional neural networks for re- source efficient inference

    Pavlo Molchanov, Stephen Tyree, Tero Karras, Timo Aila, and Jan Kautz. Pruning convolutional neural networks for re- source efficient inference. arXiv preprint arXiv:1611.06440,

    arXiv

  35. [42]

    Learning deep sketch abstraction

    Umar Riaz Muhammad, Yongxin Yang, Yi-Zhe Song, Tao Xiang, and Timothy M Hospedales. Learning deep sketch abstraction. In CVPR, 2018. 3

    2018

  36. [43]

    Goal-driven sequential data abstraction

    Umar Riaz Muhammad, Yongxin Yang, Timothy Hospedales, Tao Xiang, and Yi-Zhe Song. Goal-driven sequential data abstraction. In ICCV, 2019. 1, 3

    2019

  37. [44]

    DINOv2: Learning Robust Visual Features without Supervi- sion

    Maxime Oquab, Timoth ´ee Darcet, Th ´eo Moutakanni, Huy V o, Marc Szafraniec, Vasil Khalidov, Pierre Fernandez, Daniel Haziza, Francisco Massa, Alaaeldin El-Nouby, et al. DINOv2: Learning Robust Visual Features without Supervi- sion. arXiv preprint arXiv:2304.07193, 2023. 8

    2023 arXiv

  38. [45]

    Cross-domain generative learning for fine- grained sketch-based image retrieval

    Kaiyue Pang, Yi-Zhe Song, Tony Xiang, and Timothy M Hospedales. Cross-domain generative learning for fine- grained sketch-based image retrieval. In BMVC, 2017. 2

    2017

  39. [46]

    Gener- alising fine-grained sketch-based image retrieval

    Kaiyue Pang, Ke Li, Yongxin Yang, Honggang Zhang, Tim- othy M Hospedales, Tao Xiang, and Yi-Zhe Song. Gener- alising fine-grained sketch-based image retrieval. In CVPR,

  40. [47]

    Solving mixed-modal jigsaw puz- zle for fine-grained sketch-based image retrieval

    Kaiyue Pang, Yongxin Yang, Timothy M Hospedales, Tao Xiang, and Yi-Zhe Song. Solving mixed-modal jigsaw puz- zle for fine-grained sketch-based image retrieval. In CVPR,

  41. [48]

    Learning deep rep- resentations with probabilistic knowledge transfer

    Nikolaos Passalis and Anastasios Tefas. Learning deep rep- resentations with probabilistic knowledge transfer. InECCV,

  42. [49]

    SceneTrilogy: On Human Scene-Sketch and its Com- plementarity with Photo and Text

    Pinaki Nath Chowdhury and Ayan Kumar Bhunia and Anee- shan Sain and Subhadeep Koley and Tao Xiang and Yi-Zhe Song. SceneTrilogy: On Human Scene-Sketch and its Com- plementarity with Photo and Text. In CVPR, 2023. 2

    2023

  43. [50]

    What Can Human Sketches Do for Object Detection? In CVPR, 2023

    Pinaki Nath Chowdhury and Ayan Kumar Bhunia and Anee- shan Sain and Subhadeep Koley and Tao Xiang and Yi-Zhe Song. What Can Human Sketches Do for Object Detection? In CVPR, 2023. 1

    2023

  44. [51]

    Fitnets: Hints for thin deep nets

    Adriana Romero, Nicolas Ballas, Samira Ebrahimi Kahou, Antonie Chassang, Carlo Gatta, and Yoshua Bengio. Fitnets: Hints for thin deep nets. In ICLR, 2015. 2, 4, 6

    2015

  45. [52]

    Imagenet large scale visual recognition challenge

    Olga Russakovsky, Jia Deng, Hao Su, Jonathan Krause, San- jeev Satheesh, Sean Ma, Zhiheng Huang, Andrej Karpathy, Aditya Khosla, Michael Bernstein, et al. Imagenet large scale visual recognition challenge. IJCV, 2015. 4

    2015

  46. [53]

    Cross-modal hierarchical modelling forfine-grained sketch based image retrieval

    Aneeshan Sain, Ayan Kumar Bhunia, Yongxin Yang, Tao Xi- ang, and Yi-Zhe Song. Cross-modal hierarchical modelling forfine-grained sketch based image retrieval. In BMVC,

  47. [54]

    Stylemeup: Towards style-agnostic sketch-based image retrieval

    Aneeshan Sain, Ayan Kumar Bhunia, Yongxin Yang, Tao Xi- ang, and Yi-Zhe Song. Stylemeup: Towards style-agnostic sketch-based image retrieval. In CVPR, 2021. 1, 2, 3, 4, 6, 7, 8

    2021

  48. [55]

    Sketch3t: Test-time training for zero-shot sbir

    Aneeshan Sain, Ayan Kumar Bhunia, Vaishnav Potlapalli, Pinaki Nath Chowdhury, Tao Xiang, and Yi-Zhe Song. Sketch3t: Test-time training for zero-shot sbir. In CVPR,

  49. [57]

    Mobilenetv2: Inverted residuals and linear bottlenecks

    Mark Sandler, Andrew Howard, Menglong Zhu, Andrey Zh- moginov, and Liang-Chieh Chen. Mobilenetv2: Inverted residuals and linear bottlenecks. In CVPR, 2018. 1, 3, 4, 5, 7

    2018

  50. [58]

    The sketchy database: learning to retrieve badly drawn bunnies

    Patsorn Sangkloy, Nathan Burnell, Cusuh Ham, and James Hays. The sketchy database: learning to retrieve badly drawn bunnies. ACM TOG, 2016. 5, 6

    2016

  51. [59]

    Proximal policy optimization algo- rithms

    John Schulman, Filip Wolski, Prafulla Dhariwal, Alec Rad- ford, and Oleg Klimov. Proximal policy optimization algo- rithms. arXiv preprint arXiv:1707.06347, 2017. 5

    2017 arXiv

  52. [60]

    Very deep convo- lutional networks for large-scale image recognition

    Karen Simonyan and Andrew Zisserman. Very deep convo- lutional networks for large-scale image recognition. InICLR,

  53. [61]

    Fine-grained image retrieval: the text/sketch input dilemma

    Jifei Song, Yi-Zhe Song, Tony Xiang, and Timothy M Hospedales. Fine-grained image retrieval: the text/sketch input dilemma. In BMVC, 2017. 2

    2017

  54. [62]

    Deep spatial-semantic attention for fine- grained sketch-based image retrieval

    Jifei Song, Qian Yu, Yi-Zhe Song, Tao Xiang, and Timo- thy M Hospedales. Deep spatial-semantic attention for fine- grained sketch-based image retrieval. In ICCV, 2017. 2, 6

    2017

  55. [63]

    Picture that Sketch: Photorealistic Image Generation from Abstract Sketches

    Subhadeep Koley and Ayan Kumar Bhunia and Aneeshan Sain and Pinaki Nath Chowdhury and Tao Xiang and Yi-Zhe Song. Picture that Sketch: Photorealistic Image Generation from Abstract Sketches. In CVPR, 2023. 1

    2023

  56. [64]

    Policy gradient methods for reinforcement learning with function approximation

    Richard S Sutton, David A McAllester, Satinder P Singh, and Yishay Mansour. Policy gradient methods for reinforcement learning with function approximation. In NeurIPS, 2000. 5

    2000

  57. [65]

    Learning to resize im- ages for computer vision tasks

    Hossein Talebi and Peyman Milanfar. Learning to resize im- ages for computer vision tasks. In ICCV, 2021. 2, 6 10

    2021

  58. [66]

    Efficientnet: Rethinking model scaling for convolutional neural networks

    Mingxing Tan and Quoc Le. Efficientnet: Rethinking model scaling for convolutional neural networks. In ICML, 2019. 1, 3

    2019

  59. [67]

    Learning when and where to zoom with deep reinforcement learning

    Burak Uzkent and Stefano Ermon. Learning when and where to zoom with deep reinforcement learning. In CVPR, 2020. 2

    2020

  60. [68]

    White blood cell classification: a comparison be- tween vgg-16 and resnet-50 models

    Supawit Vatathanavaro, Suchat Tungjitnob, and Kitsuchart Pasupa. White blood cell classification: a comparison be- tween vgg-16 and resnet-50 models. In JSCI, 2018. 3

    2018

  61. [69]

    The douglas- peucker algorithm for line simplification: re-evaluation through visualization

    Mahes Visvalingam and J Duncan Whyatt. The douglas- peucker algorithm for line simplification: re-evaluation through visualization. In Computer Graphics Forum, 1990. 5

    1990

  62. [70]

    Deep reinforced attention regression for partial sketch based image retrieval

    Dingrong Wang, Hitesh Sapkota, Xumin Liu, and Qi Yu. Deep reinforced attention regression for partial sketch based image retrieval. In ICDM, 2021. 12

    2021

  63. [71]

    Reinforced cross-modal matching and self- supervised imitation learning for vision-language navigation

    Xin Wang, Qiuyuan Huang, Asli Celikyilmaz, Jianfeng Gao, Dinghan Shen, Yuan-Fang Wang, William Yang Wang, and Lei Zhang. Reinforced cross-modal matching and self- supervised imitation learning for vision-language navigation. In CVPR, 2019. 3

    2019

  64. [72]

    Cnnpack: Packing convolutional neural networks in the frequency domain

    Yunhe Wang, Chang Xu, Shan You, Dacheng Tao, and Chao Xu. Cnnpack: Packing convolutional neural networks in the frequency domain. NeurIPS, 29, 2016. 2

    2016

  65. [73]

    Glance and focus: a dynamic approach to reducing spatial redundancy in image classification

    Yulin Wang, Kangchen Lv, Rui Huang, Shiji Song, Le Yang, and Gao Huang. Glance and focus: a dynamic approach to reducing spatial redundancy in image classification. In NeurIPS, 2020. 2

    2020

  66. [74]

    Distance met- ric learning for large margin nearest neighbor classification

    Kilian Q Weinberger and Lawrence K Saul. Distance met- ric learning for large margin nearest neighbor classification. JMLR, 2009. 4

    2009

  67. [75]

    Multi- graph transformer for free-hand sketch recognition

    Peng Xu, Chaitanya K Joshi, and Xavier Bresson. Multi- graph transformer for free-hand sketch recognition. IEEE T-NNLS, 2021. 4

    2021

  68. [76]

    Deep learning for free-hand sketch: A survey

    Peng Xu, Timothy M Hospedales, Qiyue Yin, Yi-Zhe Song, Tao Xiang, and Liang Wang. Deep learning for free-hand sketch: A survey. IEEE TPAMI, 2022. 1, 2

    2022

  69. [77]

    SmartAdapt: Multi-branch object detection framework for videos on mo- biles

    Ran Xu, Fangzhou Mu, Jayoung Lee, Preeti Mukherjee, So- mali Chaterji, Saurabh Bagchi, and Yin Li. SmartAdapt: Multi-branch object detection framework for videos on mo- biles. In CVPR, 2022. 2

    2022

  70. [78]

    Hospedales

    Qian Yu, Yongxin Yang, Yi-Zhe Song, Tao Xiang, and Tim- othy M. Hospedales. Sketch-a-net that beats humans. In BMVC, 2015. 6, 8

    2015

  71. [79]

    Sketch me that shoe

    Qian Yu, Feng Liu, Yi-Zhe Song, Tao Xiang, Timothy M Hospedales, and Chen-Change Loy. Sketch me that shoe. In CVPR, 2016. 1, 2, 3, 5, 6, 7, 8

    2016

  72. [80]

    Paying more at- tention to attention: Improving the performance of convolu- tional neural networks via attention transfer

    Sergey Zagoruyko and Nikos Komodakis. Paying more at- tention to attention: Improving the performance of convolu- tional neural networks via attention transfer. In ICLR, 2017. 2, 6

    2017

  73. [81]

    Dorefa-net: Training low bitwidth convo- lutional neural networks with low bitwidth gradients

    Shuchang Zhou, Yuxin Wu, Zekun Ni, Xinyu Zhou, He Wen, and Yuheng Zou. Dorefa-net: Training low bitwidth convo- lutional neural networks with low bitwidth gradients. arXiv preprint arXiv:1606.06160, 2016. 2

    2016 arXiv

  74. [82]

    Dynamic resolution net- work

    Mingjian Zhu, Kai Han, Enhua Wu, Qiulin Zhang, Ying Nie, Zhenzhong Lan, and Yunhe Wang. Dynamic resolution net- work. In NeurIPS, 2021. 7 11 Supplementary Material for Sketch Down the FLOPs: Towards Efficient Networks for Human Sketch Further analyses: Table A. Performance & c...

    2021

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.