Deep Randomized Ensembles for Metric Learning

Xuan, Hong; Souvenir, Richard; Pless, Robert

doi:10.1007/978-3-030-01270-0_44

Cited by 94 publications

(63 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The proposed GRNet greatly suppresses the performance degradation caused by occlusions, cropping, viewpoints and small logos, outperforming existing methods with large margins as shown in Figure 2. Specifically, on the DeepFashion [34] benchmark, GRNet absolutely improves the top-1, top-20, and top-50 accuracies of the best ever reported results by 12%, 21% and 18%, and the best results of two state-of-the-art deep matching methods [43,54] 1 by 4%, 10%, and 10% respectively On Street2Shop [25] benchmark, GRNet achieves new state-of-the-art results on all five categories i.e. "tops", "dresses", "skirts", "pants" and "outerwear".…”

Section: Introductionmentioning

confidence: 94%

“…Metric learning. Our work is also related to general deep metric learning [36,56,26,37,54]. However, they only conducted experiments on InShop clothes retrieval dataset while our work focuses on customer-to-shop clothes retrieval which is much more challenging as analyzed in [34].…”

Section: Datasetsmentioning

confidence: 99%

“…We also compare GRNet with DREML [54], which achieves state-of-the-art performance on multiple general metric learning benchmarks including Inshop [34], recently. We train the DREML model on DeepFashion training set using its open source code with 192 recommended meta classes and 48 ensemble models as done in Table 2 of DREML [54]. Our GRNet is remarkably superior than DREML although DREML employs 48 models for ensemble.…”

Section: Implementation Detailsmentioning

confidence: 99%

See 2 more Smart Citations

Fashion Retrieval via Graph Reasoning Networks on a Similarity Pyramid

Kuang¹,

Gao

et al. 2019

2019 IEEE/CVF International Conference on Computer Vision (ICCV)

View full text Add to dashboard Cite

Matching clothing images from customers and online shopping stores has rich applications in E-commerce. Existing algorithms encoded an image as a global feature vector and performed retrieval with the global representation. However, discriminative local information on clothes are submerged in this global representation, resulting in suboptimal performance. To address this issue, we propose a novel Graph Reasoning Network (GRNet) on a Similarity Pyramid, which learns similarities between a query and a gallery cloth by using both global and local representations in multiple scales. The similarity pyramid is represented by a Graph of similarity, where nodes represent similarities between clothing components at different scales, and the final matching score is obtained by message passing along edges. In GRNet, graph reasoning is solved by training a graph convolutional network, enabling to align salient clothing components to improve clothing retrieval. To facilitate future researches, we introduce a new benchmark FindFashion, containing rich annotations of bounding boxes, views, occlusions, and cropping. Extensive experiments show that GRNet obtains new state-of-the-art results on two challenging benchmarks, e.g. pushing the top-1, top-20, and top-50 accuracies on DeepFashion to 26%, 64%, and 75% (i.e. 4%, 10%, and 10% absolute improvements), outperforming competitors with large margins. On FindFashion, GRNet achieves considerable improvements on all empirical settings.

show abstract

Section: Introductionmentioning

confidence: 94%

Section: Datasetsmentioning

confidence: 99%

Section: Implementation Detailsmentioning

confidence: 99%

See 1 more Smart Citation

Fashion Retrieval via Graph Reasoning Networks on a Similarity Pyramid

Kuang¹,

Gao

et al. 2019

2019 IEEE/CVF International Conference on Computer Vision (ICCV)

View full text Add to dashboard Cite

show abstract

“…Recently, several researchers in [53], [41], [1], [54] have used ensemble techniques in metric learning to boost the retrieval accuracy. Yuan et al [53] used models with different complexities obtained by networks of different depths and ensembled them in a cascade form.…”

Section: Ensemble-based Deep Metric Learningmentioning

confidence: 99%

“…The work in [1] used attention-based ensemble where different learners attend to different parts of images. Similarly, the recent work [54] used random bagging of training classes into meta-classes and trained multiple CNNs. The final embedding is then obtained by concatenating the embedding from individual networks.…”

Section: Ensemble-based Deep Metric Learningmentioning

confidence: 99%

Dynamically Modulated Deep Metric Learning for Visual Search

Manandhar

Baştan

Yap

2020

ICASSP 2020 - 2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

View full text Add to dashboard Cite

Deep metric learning applied to various multimedia applications has shown promising results in tasks such identification, retrieval and recognition. Existing metric learning methods often do not consider different granularity in visual similarity. However, in many domain applications, images exhibit similarity at multiple granularities with visual semantic concepts, e.g. fashion demonstrates similarity ranging from clothing of the exact same instance to similar looks/design or a common category. Therefore, training image triplets/pairs used for metric learning inherently possess different degree of information. However, the existing methods often treats them with equal importance during training. This hinders capturing the underlying granularities in feature similarity which is required for effective visual search.In view of this, we propose a new deep semantic granularity metric learning (SGML) that develops a novel idea of detecting and leveraging attribute semantic space to capture different granularity of similarity, and then integrate this information into deep metric learning. The proposed framework simultaneously learns image attributes and embeddings using multitask CNNs with shared parameters. The two tasks are not only jointly optimized but are further linked by the semantic granularity similarity mappings to leverage the correlations between the tasks. To this end, we propose a new soft-binomial deviance loss that effectively integrates the degree of information in training samples, which helps to capture visual similarity at multiple granularities. Compared to recent ensemble-based methods, our framework is conceptually elegant, computationally simple and provides better performance. We perform extensive experiments on benchmark metric learning datasets and demonstrate that our method outperforms recent state-of-the-art methods, e.g. , 1-4.5% improvement in Recall@1 over the previous state-of-the-arts [1], [2] on DeepFashion In-Shop dataset.

show abstract