Searcharxiv⌕ Search

arXiv subjects

Ye Du

Publications and source records attributed to Ye Du.

26 records · Page 2Linked to original sources
arXiv2022

Visual Grounding with Transformers

In this paper, we propose a transformer based approach for visual grounding. Unlike previous proposal-and-rank frameworks that rely heavily on pretrained object detectors or proposal-free frameworks that upgrade an off-the-shelf one-stage detector by fusing textual embeddings, our approach is built on top of a transformer encoder-decoder and is independent of any pretrained detectors or word embedding models. Termed VGTR -- Visual Grounding with TRansformers, our approach is designed to learn semantic-discriminative visual features under the guidance of the textual description without harming their location ability. This information flow enables our VGTR to have a strong capability in capturing context-level semantics of both vision and language modalities, rendering us to aggregate accurate visual clues implied by the description to locate the interested object instance. Experiments show that our method outperforms state-of-the-art proposal-free approaches by a considerable margin on five benchmarks while maintaining fast inference speed.

cs.CV↗
arXiv2022

Weakly Supervised Semantic Segmentation by Pixel-to-Prototype Contrast

Though image-level weakly supervised semantic segmentation (WSSS) has achieved great progress with Class Activation Maps (CAMs) as the cornerstone, the large supervision gap between classification and segmentation still hampers the model to generate more complete and precise pseudo masks for segmentation. In this study, we propose weakly-supervised pixel-to-prototype contrast that can provide pixel-level supervisory signals to narrow the gap. Guided by two intuitive priors, our method is executed across different views and within per single view of an image, aiming to impose cross-view feature semantic consistency regularization and facilitate intra(inter)-class compactness(dispersion) of the feature space. Our method can be seamlessly incorporated into existing WSSS models without any changes to the base networks and does not incur any extra inference burden. Extensive experiments manifest that our method consistently improves two strong baselines by large margins, demonstrating the effectiveness. Specifically, built on top of SEAM, we improve the initial seed mIoU on PASCAL VOC 2012 from 55.4% to 61.5%. Moreover, armed with our method, we increase the segmentation mIoU of EPS from 70.8% to 73.6%, achieving new state-of-the-art.

cs.CV↗
arXiv2021

ARM: A Confidence-Based Adversarial Reweighting Module for Coarse Semantic Segmentation

Coarsely-labeled semantic segmentation annotations are easy to obtain, but therefore bear the risk of losing edge details and introducing background pixels. Impeded by the inherent noise, existing coarse annotations are only taken as a bonus for model pre-training. In this paper, we try to exploit their potentials with a confidence-based reweighting strategy. To expand, loss-based reweighting strategies usually take the high loss value to identify two completely different types of pixels, namely, valuable pixels in noise-free annotations and mislabeled pixels in noisy annotations. This makes it impossible to perform two tasks of mining valuable pixels and suppressing mislabeled pixels at the same time. However, with the help of the prediction confidence, we successfully solve this dilemma and simultaneously perform two subtasks with a single reweighting strategy. Furthermore, we generalize this strategy into an Adversarial Reweighting Module (ARM) and prove its convergence strictly. Experiments on standard datasets shows our ARM can bring consistent improvements for both coarse annotations and fine annotations. Specifically, built on top of DeepLabv3+, ARM improves the mIoU on the coarsely-labeled Cityscapes by a considerable margin and increases the mIoU on the ADE20K dataset to 47.50.

cs.CV↗
arXiv2021

Spin-current-related magnetoresistance in epitaxial Pt/Co bilayers in the presence of spin Hall effect and Rashba-Edelstein effect

We analyze the experimentally obtained spin-current-related magnetoresistance in epitaxial Pt/Co bilayers by using a drift-diffusion model that incorporates both bulk spin Hall effect and interfacial Rashba-Edelstein effect (REE). The magnetoresistance analysis yields, for the Pt/Co interface, a temperature-independent Rashba parameter in the order of 1e-11 eV m that agrees with theoretical calculations, along with an effective interfacial REE thickness of several angstroms which is in overall consistency with our previous spin-orbit torque analysis. In particular, our results suggest that both bulk and interface charge-spin current inter-conversions need to be taken into account for the spin-current-related magnetoresistance analysis in highly-conductive magnetic hetero-structures such as the epitaxial Pt/Co bilayers.

cond-mat.mtrl-sci↗
arXiv2014

The Discrete Sell or Hold Problem with Constraints on Asset Values

The discrete sell or hold problem (DSHP), which is introduced in \cite{H12}, is studied under the constraint that each asset can only take a constant number of different values. We show that if each asset can take only two values, the problem becomes polynomial-time solvable. However, even if each asset can take three different values, DSHP is still NP-hard. An approximation algorithm is also given under this setting.

cs.GT↗
arXiv2009

Ranking via Arrow-Debreu Equilibrium

In this paper, we establish a connection between ranking theory and general equilibrium theory. First of all, we show that the ranking vector of PageRank or Invariant method is precisely the equilibrium of a special Cobb-Douglas market. This gives a natural economic interpretation for the PageRank or Invariant method. Furthermore, we propose a new ranking method, the CES ranking, which is minimally fair, strictly monotone and invariant to reference intensity, but not uniform or weakly additive.

cs.GT↗