Pith. sign in

REVIEW 3 major objections 7 minor 37 references

Learning a Unified Embedding for Visual Search at Pinterest

T0 review · 3 major / 7 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read A single jointly trained image embedding can serve Pinterest's browsing, camera, and shopping visual search products, outperforming specialized embeddings on every product.

desk verdict A solid industrial multi-task embedding paper: the deployment claim is credible, but the specific claim that multi-task training alone drives the gains isn't fully isolated from architecture and data confounds. read the letter →

arxiv 1908.01707 v1 pith:VEMIMV3E submitted 2019-08-05 cs.CV

Andrew Zhai , Hao-Yu Wu , Eric Tzeng , Dong Huk Park , Charles Rosenberg This is my paper · ORCID
classification cs.CV
keywords multi-taskmetriclearningvisualsearchimageembeddingbinarizationsubsamplingengagementdataretrievalPinterest
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 show that a single image embedding, trained jointly for three different visual search tasks, can replace the separate specialized embeddings Pinterest used for each product. That matters because maintaining one embedding instead of three cuts storage, back-fill, and engineering costs while improving relevance and engagement. The authors argue that multi-task training lets the model exploit correlations across datasets—web images, camera photos, and clean product shots—so that each product gets better than it would from training alone. They support this with offline retrieval metrics, human relevance judgements, and online A/B tests, and note that the embedding can be binarized to binary codes with little quality loss.

What carries the argument

The mechanism is classification-based metric learning extended to multiple tasks: a shared base network (SE-ResNeXt101) outputs an embedding, and each task attaches its own fully-connected softmax layer whose weight columns act as class proxies; all tasks are trained with equal-weighted cross-entropy losses on uniformly sampled mini-batches. Two components make this practical at web scale: a subsampling module keeps the proxy bank in CPU memory and optimizes only a random subset of classes per batch, and a binarization module (GroupNorm, ReLU, dropout) produces binary embeddings that preserve most of the float performance. The proxy view is what lets many tasks with different label structures share one embedding without pairwise sampling.

What would settle it

Retrain the old specialized embeddings with the same SE-ResNeXt101 architecture and evaluate them on the exact Table 4 corpora; if the unified embedding no longer beats them, the 'unified beats specialized' claim is explained by the backbone change. Conversely, hold out a human-labeled relevance set drawn from live query traffic for Flashlight and Lens and check whether the multi-task model still beats single-dataset models; if the gap vanishes outside the training class distribution, the offline improvements are an artifact of label reuse.

Watch

Extended reading notes

Core claim

The central discovery is that learning one embedding by jointly optimizing four softmax classification objectives—Flashlight class, Shop-the-Look product class, Shop-the-Look instance class, and Lens category class—improves all three retrieval products compared with same-architecture embeddings trained on each dataset alone and compared with the older specialized embeddings. The shared network captures general visual structure while each task branch keeps its own proxy classifiers, and the interaction of the datasets helps each task: for example, Shop-the-Look crops teach the embedding to handle cropped inputs, and Flashlight engagement data organizes Pinterest content that Lens needs. The paper's conclusion is that a unified embedding is not just an engineering simplification but a quality improvement.

Load-bearing premise

The load-bearing premise is that engagement-derived labels (clicks, saves, and related-pin patterns) truly capture visual relevance; if those labels are noisy or biased for the Flashlight and Lens tasks, the offline multi-task gains may reflect the label structure rather than better retrieval, and only the human-judgement and A/B results would remain as evidence.

Editorial extensions

If this is right

  • Pinterest can replace its three embedding lineages with one model, reducing back-fill, dependency, and serving costs without sacrificing quality.
  • Because the learned embedding can be binarized, billions of images can be stored and compared efficiently as binary codes, which keeps retrieval cheap as the corpus grows.
  • Multi-task training converts dataset choice into a model-learned decision: the model uses correlated information from all datasets, so each product improves beyond what its own labels alone provide.
  • The same four-task objective is the deployment model: Flashlight, Lens, and Shop-the-Look can all be served from one embedding with separate ranking layers on top.

Reading between the lines

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

  • Editorial inference: Because the unified model uses a newer SE-ResNeXt101 backbone while the old specialized embeddings use VGG16, ResNeXt50, and ResNet101, part of the measured improvement in Table 4 likely comes from the architecture update rather than from multi-task learning itself; a same-backbone comparison would separate the two effects.
  • Editorial inference: The offline Flashlight and Lens evaluations draw on the same engagement-derived label structure used for training, so those particular numbers may overstate real-world relevance; the human-judgement and A/B results are therefore the stronger evidence for the paper's claim.
  • Editorial inference: The same recipe—proxy classification, subsampling, binarization, and equal-weighted multi-task losses—could plausibly extend to other embedding consumers at Pinterest, such as near-duplicate detection or pin-to-pin ranking, provided those tasks can be expressed as class-proxy objectives; the paper does not test this.
  • Editorial inference: If the unified embedding generalizes as claimed, a single embedding space can serve as shared infrastructure for both browsing and shopping, which suggests that other platforms with multiple visual search products could collapse their separate systems similarly; this is speculative beyond the paper's data.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 7 minor

Summary. The paper describes a multi-task deep metric learning system that learns a single unified image embedding for three Pinterest visual search products: Flashlight (browse from web images), Lens (browse from camera photos), and Shop-the-Look (search for exact products). The method extends a classification-based proxy metric learning approach with a shared SE-ResNeXt101 backbone, task-specific proxy heads, class subsampling for scalability, and a GroupNorm-based binarization module. Joint training balances the three datasets by uniform mini-batch mixing and equal loss weights, with sparse tensor optimization. The evaluation includes offline retrieval metrics on product-specific datasets, human relevance judgements, and online A/B experiments for Flashlight and Lens. The paper claims the unified embedding outperforms all previously deployed specialized embeddings and that multi-task training improves all three retrieval metrics compared to single-task training. Deployment at Pinterest reportedly reduced engineering and operational costs while improving quality.

Significance. If the results hold, this is a significant industrial demonstration that a single multi-task embedding can replace several specialized visual-search embeddings, reducing maintenance and storage costs while improving relevance and engagement. The paper provides a concrete architecture, a scalable subsampling strategy, and a binarization method that are directly applicable to web-scale systems. The evidence base is broad: offline ablations, human judgements, and live A/B experiments, with detailed descriptions of training data and deployment. The honest discussion of fragmented legacy architectures and the explicit acknowledgement of noisy engagement-derived labels are strengths. However, the central causal claim that multi-task training itself improves all three metrics rests on offline evaluations whose independence from the training labels is not established, and the human/A-B comparisons confound multi-task learning with architecture changes.

major comments (3)
  1. [Section 4.2, Table 3] The offline evaluations for Flashlight and Lens are not independent of the engagement-derived labels used in training. The Flashlight evaluation is explicitly described as 'a random class subset of the Flashlight training data' (Section 4.2), and the Lens evaluation is 'generated using user engagement in the same manner as the Flashlight training dataset.' Since the model is trained to classify exactly these class structures, the retrieval metric measures how well the model fits the training label distribution rather than visual relevance on held-out classes. This circularity is load-bearing for the Section 4.2.3 claim that 'multi-task improves all three retrieval metrics,' because that claim is based solely on Table 3. Please provide an evaluation set with disjoint classes or an independently human-labeled retrieval benchmark for Flashlight and Lens, or explicitly reframe the Table 3 results as a measure of fit to the engagement label structure.
  2. [Section 4.3-4.4, Tables 5-7] The human judgement and A/B experiments compare the unified model (multi-task, SE-ResNeXt101, all three datasets) against the legacy specialized embeddings, which use different backbones (VGG16, ResNeXt50, ResNet101) and were trained on single datasets. These experiments therefore cannot isolate the contribution of multi-task learning; they demonstrate that the new unified system outperforms the old production systems, but architecture and data changes are confounded with the multi-task objective. The paper's specific causal claim about multi-task benefit thus rests on the offline ablation in Table 3, which suffers from the independence problem noted above. To make the multi-task claim robust, I request either a human/A-B comparison that holds architecture fixed between single-task and multi-task training, or a clear statement that the human/A-B results support the unified system without isolating the multi-task effect.
  3. [Section 3.4.2] The momentum approximation for sparse tensors is a nontrivial optimization change: the authors replace momentum updates by increasing the learning rate 10x, justified by a geometric-series argument. The text states this 'retains comparable performance' but no quantitative ablation is provided. Since this modification affects the optimization trajectory and is part of the method's scalability claims, please add a table or figure comparing convergence and final metrics with and without this approximation.
minor comments (7)
  1. [Section 3.2.3] The phrase '189 product class label' should be '189 product class labels' or '189 product classes' for grammatical correctness.
  2. [Section 3.3.1, Algorithm 1] The while-loop condition 'len(sampled_proxy_idx) <= num_samples' appears to permit sampling one more proxy than num_samples; if num_samples is an exact maximum, the condition should be '<'. The inner enumeration over sampled_proxy_idx is O(M*N), which is fine for the described scale but worth a complexity note.
  3. [Section 4.2.1, Table 1] The caption notation 'f = float, b = binary' is terse; it would be clearer to state explicitly that all rows after the first row are binary embeddings, and that the first two rows are float vs binary versions of the baseline.
  4. [Section 4.2] There is a typo in the text: 'Precison@1' should be 'Precision@1'.
  5. [Section 4.2] The Shop-the-Look offline evaluation is small (600 objects, 1421 ground-truth matches); please note this limitation in the text and consider reporting confidence intervals for P@1.
  6. [Section 4.4] It is stated that Shop-the-Look had not launched, so no A/B experiment was run. This is acceptable, but the paper should explicitly note that the Shop-the-Look claim rests on human judgement only, not on online engagement measurements.
  7. [References] Reference [11] is missing a year and full publication details; please complete the bibliographic information.

Circularity Check

1 steps flagged · score 6.0 of 10

The multi-task improvement claim in Table 3 rests on offline Flashlight/Lens evaluations built from the same engagement labels used for training, so this specific prediction is partially circular; human/A-B tests do not isolate multi-task.

  1. fitted input called prediction [Section 4.2 (Offline Evaluation), especially 'Multi-Task Dataset Ablations' Section 4.2.3 and Table 3]
    "The evaluation datasets for Flashlight and Lens are generated through user engagement. For Flashlight, we sampled a random class subset of the Flashlight training data (Section 3.2), and randomly divided it into 807 images for queries and 42881 images for the corpus across classes. For Lens, we generated the evaluation dataset using user engagement in the same manner as the Flashlight training dataset (Section 3.2) but filtering the query images to be camera images with human judgement."

    The Flashlight AP@20 metric is computed on images and classes drawn from the Flashlight training set, and the Lens AP@20 metric uses labels produced by the same engagement heuristics that generated Flashlight training labels. Table 3 then compares models trained on those very labels and reports that multi-task 'improves all three retrieval metrics.' The offline prediction is therefore not an independent measure of relevance: it measures fit to the same engagement-derived class structure the model was optimized against. Human judgements and A/B experiments do compare the unified deployment against old specialized embeddings, but those systems differ in architecture and data, so they do not isolate the multi-task-training contribution that Section 4.2.3 asserts.

full rationale

The paper's deployment-level claim (unified embedding beats old specialized embeddings) has genuinely independent support from human judgement studies (Tables 5-6) and A/B experiments (Figure 6, Table 7), so the paper is not wholly circular. However, the specific and load-bearing claim that joint multi-task training itself improves all three retrieval metrics is supported only by Table 3, whose Flashlight and Lens evaluations are built from the same engagement-derived labels and, for Flashlight, from a subset of the training data. That is a fitted-input evaluation rather than an external check. The human and online experiments compare the full unified system (new architecture plus all data) against older specialized systems, so they cannot rescue the isolated multi-task attribution. Self-citations to the authors' prior classification-based metric learning work [33] are used as architectural scaffolding but are not the source of the multi-task claim, so they do not add circularity. Overall: one partial reduction of the offline 'prediction' to its training labels, with independent evidence for the broader product claim, warrants a score of 6 rather than 0-2.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The central claim rests on domain assumptions (engagement as relevance, proxy classification as a retrieval surrogate, zero-threshold binarization) and several hand-chosen hyperparameters. These are reasonable for an applied system paper, but they mean the result is an empirical demonstration conditional on Pinterest's data and label definitions rather than a parameter-free derivation.

free parameters (5)
  • NormSoftmax temperature tau = 0.05
    Inherited from prior work [33]; chosen by hand; controls embedding concentration and is not fitted to the multi-task data.
  • Subsampled proxy count = 2048
    Chosen as a compute and memory tradeoff; softmax random-guessing loss floor ln(2048) is noted in Section 3.3.1.
  • GroupNorm group count = 256
    Chosen in the binarization module for granularity of normalization, Section 4.2.1.
  • Dropout probability = 0.5
    Used before binarization to learn redundant representations, Section 4.2.1.
  • Equal task loss weights = 1.0 each
    Chosen over dataset-size sampling and GradNorm after offline comparison, Table 2; performance depends on this choice.
assumptions (4)
  • domain assumption Classification with proxies is a valid surrogate for retrieval relevance
    Adopted from the authors' prior work [33] and used throughout the method section (Section 2.2).
  • domain assumption User engagement signals (closeups, repins) approximate visual relevance
    Used to generate the Flashlight and Lens training datasets and offline evaluations, Sections 3.2.1 and 4.2.
  • domain assumption Zero-threshold binarization preserves retrieval quality
    Assumed from prior work [33] and verified only internally in Table 1, Section 3.3.2.
  • domain assumption Uniform dataset mixing with equal loss weights is near-optimal for this system
    Empirically tested against dataset-size sampling and GradNorm in Section 4.2.2, but no theoretical justification.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Learning a Unified Embedding for Visual Search at Pinterest." pith.science (2026). https://pith.science/paper/VEMIMV3E

@misc{pith2026190801707,
  author       = {Pith},
  title        = {Pith review of: Learning a Unified Embedding for Visual Search at Pinterest},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VEMIMV3E}},
  note         = {Machine review of arXiv:1908.01707}
}
read the original abstract

At Pinterest, we utilize image embeddings throughout our search and recommendation systems to help our users navigate through visual content by powering experiences like browsing of related content and searching for exact products for shopping. In this work we describe a multi-task deep metric learning system to learn a single unified image embedding which can be used to power our multiple visual search products. The solution we present not only allows us to train for multiple application objectives in a single deep neural network architecture, but takes advantage of correlated information in the combination of all training data from each application to generate a unified embedding that outperforms all specialized embeddings previously deployed for each product. We discuss the challenges of handling images from different domains such as camera photos, high quality web images, and clean product catalog images. We also detail how to jointly train for multiple product objectives and how to leverage both engagement data and human labeled data. In addition, our trained embeddings can also be binarized for efficient storage and retrieval without compromising precision and recall. Through comprehensive evaluations on offline metrics, user studies, and online A/B experiments, we demonstrate that our proposed unified embedding improves both relevance and engagement of our visual search products for both browsing and searching purposes when compared to existing specialized embeddings. Finally, the deployment of the unified embedding at Pinterest has drastically reduced the operation and engineering cost of maintaining multiple embeddings while improving quality.

Figures

Figures reproduced from arXiv: 1908.01707 by the authors.

Figure 1
Figure 1. Three visual search products on Pinterest (Flashlight, Lens, and Shop-the-Look) allowing users to browse content [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. 3.2.1 Flashlight Dataset. The Flashlight dataset is collected to de￾fine browse relevance for Pinterest (web) images. As a surrogate for relevance, we rely on engagement through feedback from our users in a similar manner to Memboost described in [14] to generate the dataset. Image sets are collected where a given query image has a set of related images ranked via engagement, and we assign each [PITH_FULL_IMAGE:fig… view at source ↗
Figure 3
Figure 3. The overall architecture for multi-task metric learning network. The proposed classification network as proxy-based [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Visualization of the subsampling module. [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: Human Judgement Task templates for Flashlight [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: A/B experiment results on Flashlight showing [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]
Figure 7
Figure 7. Figure 7: Qualitative results comparing the old embeddings [PITH_FULL_IMAGE:figures/full_fig_p009_7.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

37 extracted references · 19 canonical work pages

  1. [33]

    Andrew Zhai and Hao-Yu Wu. 2018. Making Classification Competitive for Deep Metric Learning. CoRR abs/1811.12649 (2018). arXiv:1811.12649 http: //arxiv.org/abs/1811.12649

    arXiv 2018

  2. [1]

    Sean Bell and Kavita Bala. 2015. Learning Visual Similarity for Product Design with Convolutional Neural Networks. ACM Trans. on Graphics (SIGGRAPH) 34, 4 (2015)

    work page 2015

  3. [2]

    Jerry Zitao Liu Yuchen Liu Rahul Sharma Charles Sugnet Mark Ulrich Jure Leskovec Chantat Eksombatchai, Pranav Jindal. 2018. Pixie: A System for Recommending 3+ Billion Items to 200+ Million Users in Real-Time. In Pro- ceedings of the International Conference on World Wide Web

    work page 2018

  4. [3]

    Zhao Chen, Vijay Badrinarayanan, Chen-Yu Lee, and Andrew Rabinovich

    work page

  5. [4]

    Sumit Chopra, Raia Hadsell, and Yann LeCun. 2005. Learning a Similarity Metric Discriminatively, with Application to Face Verification. In2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR’05) , Vol. 1. IEEE, 539–546

    work page 2005

  6. [5]

    J. Deng, W. Dong, R. Socher, L.-J. Li, K. Li, and L. Fei-Fei. 2009. ImageNet: A Large-Scale Hierarchical Image Database. In CVPR09

    work page 2009

  7. [6]

    Girshick, Pieter Noordhuis, Lukasz Wesolowski, Aapo Kyrola, Andrew Tulloch, Yangqing Jia, and Kaiming He

    Priya Goyal, Piotr Dollár, Ross B. Girshick, Pieter Noordhuis, Lukasz Wesolowski, Aapo Kyrola, Andrew Tulloch, Yangqing Jia, and Kaiming He. 2017. Accurate, Large Minibatch SGD: Training ImageNet in 1 Hour. CoRR abs/1706.02677 (2017). arXiv:1706.02677 http://arxiv.org/abs/1706.02677

    arXiv 2017

  8. [7]

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. 2015. Deep Residual Learning for Image Recognition. arXiv preprint arXiv:1512.03385 (2015)

    arXiv 2015

Show all 37 references
  1. [8]

    Elad Hoffer and Nir Ailon. 2014. Deep metric learning using Triplet network. CoRR abs/1412.6622 (2014). http://arxiv.org/abs/1412.6622

    2014 arXiv

  2. [9]

    Houdong Hu, Yan Wang, Linjun Yang, Pavel Komlev, Li Huang, Xi (Stephen) Chen, Jiapei Huang, Ye Wu, Meenaz Merchant, and Arun Sacheti. 2018. Web-Scale Responsive Visual Search at Bing. In Proceedings of the 24th ACM SIGKDD Inter- national Conference on Knowledge Discovery & Dat...

    2018

  3. [10]

    Jie Hu, Li Shen, and Gang Sun. 2017. Squeeze-and-excitation networks. arXiv preprint arXiv:1709.01507 (2017)

    2017 arXiv

  4. [11]

    Y. Jing, D. Liu, D. Kislyuk, A. Zhai, J. Xu, and J. Donahue. [n. d.]. Visual Search at Pinterest. In Proceedings of the International Conference on Knowledge Discovery and Data Mining (SIGKDD)

  5. [12]

    Alex Kendall, Yarin Gal, and Roberto Cipolla. 2017. Multi-Task Learning Us- ing Uncertainty to Weigh Losses for Scene Geometry and Semantics. CoRR abs/1705.07115 (2017)

    2017 arXiv

  6. [13]

    Krizhevsky, S

    A. Krizhevsky, S. Ilya, and G. E. Hinton. 2012. ImageNet Classification with Deep Convolutional Neural Networks. In Advances in Neural Information Processing Systems (NIPS). 1097–1105

    2012

  7. [14]

    Liu, Stephanie Rogers, Raymond Shiau, Dmitry Kislyuk, Kevin C

    David C. Liu, Stephanie Rogers, Raymond Shiau, Dmitry Kislyuk, Kevin C. Ma, Zhigang Zhong, Jenny Liu, and Yushi Jing. 2017. Related Pins at Pinterest: The Evolution of a Real-World Recommender System. CoRR abs/1702.07969 (2017)

    2017 arXiv

  8. [15]

    Ishan Misra, Abhinav Shrivastava, Abhinav Gupta, and Martial Hebert. 2016. Cross-stitch Networks for Multi-task Learning. CoRR abs/1604.03539 (2016)

    2016 arXiv

  9. [16]

    Leung, Sergey Ioffe, and Saurabh Singh

    Yair Movshovitz-Attias, Alexander Toshev, Thomas K. Leung, Sergey Ioffe, and Saurabh Singh. 2017. No Fuss Distance Metric Learning using Proxies. CoRR abs/1703.07464 (2017). http://arxiv.org/abs/1703.07464

    2017 arXiv

  10. [17]

    Squire, Stéphane Marchand- Maillet, and Thierry Pun

    Henning Müller, Wolfgang Müller, David McG. Squire, Stéphane Marchand- Maillet, and Thierry Pun. 2001. Performance Evaluation in Content-based Image Retrieval: Overview and Proposals. Pattern Recogn. Lett. 22, 5 (April 2001), 593–

    2001

  11. [18]

    Zhongzheng Ren and Yong Jae Lee. 2017. Cross-Domain Self-supervised Multi- task Feature Learning using Synthetic Imagery. CoRR abs/1711.09082 (2017)

    2017 arXiv

  12. [19]

    Hamilton Jure Leskovec Rex Ying, Ruining He

    Kaifeng Chen Pong Eksombatchai William L. Hamilton Jure Leskovec Rex Ying, Ruining He. 2018. Graph Convolutional Neural Networks for Web-Scale Rec- ommender Systems. In Proceedings of the International Conference on Knowledge Discovery and Data Mining (SIGKDD)

    2018

  13. [20]

    Florian Schroff, Dmitry Kalenichenko, and James Philbin. 2015. FaceNet: A Unified Embedding for Face Recognition and Clustering. In The IEEE Conference on Computer Vision and Pattern Recognition (CVPR)

    2015

  14. [21]

    Simonyan and A

    K. Simonyan and A. Zisserman. 2014. Very Deep Convolutional Networks for Large-Scale Image Recognition. CoRR abs/1409.1556 (2014)

    2014 arXiv

  15. [22]

    Kihyuk Sohn. 2016. Improved Deep Metric Learning with Multi-class N-pair Loss Objective. In Advances in Neural Information Processing Systems 29 , D. D. Lee, M. Sugiyama, U. V. Luxburg, I. Guyon, and R. Garnett (Eds.). Curran Associates, Inc., 1857–1865

    2016

  16. [23]

    Hyun Oh Song, Yu Xiang, Stefanie Jegelka, and Silvio Savarese. 2016. Deep Metric Learning via Lifted Structured Feature Embedding. In The IEEE Conference on Computer Vision and Pattern Recognition (CVPR)

    2016

  17. [24]

    Belongie, and Theofanis Karaletsos

    Andreas Veit, Serge J. Belongie, and Theofanis Karaletsos. 2017. Conditional Similarity Networks. 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2017), 1781–1789

    2017

  18. [25]

    Bin Yang Wenjie Luo and Raquel Urtasun. 2018. Fast and Furious: Real Time End-to-End 3D Detection, Tracking and Motion Forecasting with a Single Convo- lutional Net. In The IEEE Conference on Computer Vision and Pattern Recognition (CVPR)

    2018

  19. [26]

    Manmatha, Alexander J

    Chao-Yuan Wu, R. Manmatha, Alexander J. Smola, and Philipp Krähenbühl. 2017. Sampling Matters in Deep Embedding Learning. CoRR abs/1706.07567 (2017). arXiv:1706.07567 http://arxiv.org/abs/1706.07567

    2017 arXiv

  20. [27]

    Yuxin Wu and Kaiming He. 2018. Group Normalization. CoRR abs/1803.08494 (2018)

    2018 arXiv

  21. [28]

    Saining Xie, Ross Girshick, Piotr Dollár, Zhuowen Tu, and Kaiming He. 2017. Aggregated residual transformations for deep neural networks. In 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) . IEEE, 5987–5995

    2017

  22. [29]

    Yamaguchi, M

    K. Yamaguchi, M. H. Kiapour, and T. L. Berg. 2013. Paper Doll Parsing: Retrieving Similar Styles to Parse Clothing Items. In 2013 IEEE International Conference on Computer Vision (ICCV), Vol. 00. 3519–3526. https://doi.org/10.1109/ICCV.2013. 437

    2013 doi

  23. [30]

    Hadi Kiapour, and Robinson Piramuthu

    Fan Yang, Ajinkya Kale, Yury Bubnov, Leon Stein, Qiaosong Wang, M. Hadi Kiapour, and Robinson Piramuthu. 2017. Visual Search at eBay. In Proceedings of the 23rd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, Halifax, NS, Canada, August 13 - 17, 201...

    2017

  24. [31]

    Shen, Leonidas J

    Amir Roshan Zamir, Alexander Sax, William B. Shen, Leonidas J. Guibas, Jitendra Malik, and Silvio Savarese. 2018. Taskonomy: Disentangling Task Transfer Learning. CoRR abs/1804.08328 (2018)

    2018 arXiv

  25. [32]

    Andrew Zhai, Dmitry Kislyuk, Yushi Jing, Michael Feng, Eric Tzeng, Jeff Donahue, Yue Li Du, and Trevor Darrell. 2017. Visual Discovery at Pinterest. arXiv preprint arXiv:1702.04680 (2017)

    2017 arXiv

  26. [34]

    Yanhao Zhang, Pan Pan, Yun Zheng, Kang Zhao, Yingya Zhang, Xiaofeng Ren, and Rong Jin. 2018. Visual Search at Alibaba. In Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, KDD 2018, London, UK, August 19-23, 2018 . 993–1001. http...

    2018 doi

  27. [35]

    Xiangyun Zhao, Haoxiang Li, Xiaohui Shen, Xiaodan Liang, and Ying Wu. 2018. A Modulation Module for Multi-task Learning with Applications in Image Retrieval. In ECCV

    2018

  28. [601]

    https://doi.org/10.1016/S0167-8655(00)00118-5

    doi

  29. [2017]

    CoRR abs/1711.02257 (2017)

    GradNorm: Gradient Normalization for Adaptive Loss Balancing in Deep Multitask Networks. CoRR abs/1711.02257 (2017). arXiv:1711.02257 http://arxiv.org/abs/1711.02257

    2017 arXiv

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.