Search

We propose a conceptually simple and thus fast multi-object tracking (MOT) model that does not require any attached modules, such as the Kalman filter, Hungarian algorithm, transformer blocks, or graph networks. Conventional MOT models are built upon the multi-step modules listed above, and thus the computational cost is high. Our proposed end-to-end MOT model, \textit{TicrossNet}, is composed of a base detector and a cross-attention module only. As a result, the overhead of tracking does not increase significantly even when the number of instances ($N_t$) increases. We show that TicrossNet runs \textit{in real-time}; specifically, it achieves 32.6 FPS on MOT17 and 31.0 FPS on MOT20 (Tesla V100), which includes as many as $>$100 instances per frame. We also demonstrate that TicrossNet is robust to $N_t$; thus, it does not have to change the size of the base detector, depending on $N_t$, as is often done by other models for real-time processing., Comment: Accepted to IEEE International Conference on Image Processing (IEEE ICIP) 2023

Owing to its unique geometry, the kagome lattice hosts various many-body quantum states including frustrated magnetism, superconductivity, and charge-density waves (CDWs), with intense efforts focused on kagome metals exhibiting $2\times2$ CDWs associated with the nesting of van Hove saddle points. Recently, a $\sqrt{3}\times\sqrt{3}$ CDW was discovered in the kagome metal ScV$_6$Sn$_6$ below $T_{\rm CDW}\approx91$~K, whose underlying mechanism and formation process remain unclear. Using inelastic X-ray scattering, we discover a short-range $\sqrt{3}\times\sqrt{3}\times2$ CDW that is dominant in ScV$_6$Sn$_6$ well above $T_{\rm CDW}$, distinct from the $\sqrt{3}\times\sqrt{3}\times3$ CDW below $T_{\rm CDW}$. The short-range CDW grows upon cooling, and is accompanied by the softening of phonons, indicative of its dynamic nature. As the $\sqrt{3}\times\sqrt{3}\times3$ CDW appears, the short-range CDW becomes suppressed, revealing a competition between these CDW instabilities. Our first-principles calculations indicate that the $\sqrt{3}\times\sqrt{3}\times2$ CDW is energetically favored, consistent with experimental observations at high temperatures. However, the $\sqrt{3}\times\sqrt{3}\times3$ CDW is selected as the ground state likely due to a large wavevector-dependent electron-phonon coupling, which also accounts for the enhanced electron scattering above $T_{\rm CDW}$. The competing CDW instabilities in ScV$_6$Sn$_6$ lead to an unusual CDW formation process, with the most pronounced phonon softening and the static CDW occurring at different wavevectors., Comment: Supplementary Information available upon request

Robot navigation with deep reinforcement learning (RL) achieves higher performance and performs well under complex environment. Meanwhile, the interpretation of the decision-making of deep RL models becomes a critical problem for more safety and reliability of autonomous robots. In this paper, we propose a visual explanation method based on an attention branch for deep RL models. We connect attention branch with pre-trained deep RL model and the attention branch is trained by using the selected action by the trained deep RL model as a correct label in a supervised learning manner. Because the attention branch is trained to output the same result as the deep RL model, the obtained attention maps are corresponding to the agent action with higher interpretability. Experimental results with robot navigation task show that the proposed method can generate interpretable attention maps for a visual explanation., Comment: 8 pages, 8 figures, 1 table

We have measured the lattice volume of ice VIII in different pressure-temperature pathways and found isothermal compression at low-temperature conditions makes the volume larger. Ice VIII has become its high-pressure phase with the molar volume of 6.45 cm^3 at 10 K where the pressure can be estimated as 60.4 GPa based on the third-order Birch-Murnaghan equation with parameters determined in this study (K_0= 32.4 GPa, K'_0= 3.7, and V_0= 11.9 cm^3). The present results indicate that this high-pressure state is paraelectric with tetragonal symmetry., Comment: 16 pages, 2 tables, and 5 figures with supplementary information containing 3 tables and 4 figures

An accurate pressure scale is a fundamental requirement to understand planetary interiors. Here, we establish a primary pressure scale extending to the multi-megabar pressures of the Earth's core, by combined measurement of the acoustic velocities and the density from a rhenium sample in a diamond anvil cell using inelastic x-ray scattering and x-ray diffraction. Our scale agrees well with previous primary scales and shock Hugoniots in each experimental pressure range, and reveals that previous scales have overestimated laboratory pressures by at least 20% at 230 gigapascals. It suggests that the light element content in the Earth's inner core (the density deficit relative to iron) is likely to be double what was previously estimated, or the Earth's inner core temperature is much higher than expected, or some combination thereof.

In this work, we aim to realize a method for embedding human knowledge into deep neural networks. While the conventional method to embed human knowledge has been applied for non-deep machine learning, it is challenging to apply it for deep learning models due to the enormous number of model parameters. To tackle this problem, we focus on the attention mechanism of an attention branch network (ABN). In this paper, we propose a fine-tuning method that utilizes a single-channel attention map which is manually edited by a human expert. Our fine-tuning method can train a network so that the output attention map corresponds to the edited ones. As a result, the fine-tuned network can output an attention map that takes into account human knowledge. Experimental results with ImageNet, CUB-200-2010, and IDRiD demonstrate that it is possible to obtain a clear attention map for a visual explanation and improve the classification performance. Our findings can be a novel framework for optimizing networks through human intuitive editing via a visual interface and suggest new possibilities for human-machine cooperation in addition to the improvement of visual explanations., Comment: 10 pages, 10 figures

Visual explanation enables human to understand the decision making of Deep Convolutional Neural Network (CNN), but it is insufficient to contribute the performance improvement. In this paper, we focus on the attention map for visual explanation, which represents high response value as the important region in image recognition. This region significantly improves the performance of CNN by introducing an attention mechanism that focuses on a specific region in an image. In this work, we propose Attention Branch Network (ABN), which extends the top-down visual explanation model by introducing a branch structure with an attention mechanism. ABN can be applicable to several image recognition tasks by introducing a branch for attention mechanism and is trainable for the visual explanation and image recognition in end-to-end manner. We evaluate ABN on several image recognition tasks such as image classification, fine-grained recognition, and multiple facial attributes recognition. Experimental results show that ABN can outperform the accuracy of baseline models on these image recognition tasks while generating an attention map for visual explanation. Our code is available at https://github.com/machine-perception-robotics-group/attention_branch_network., Comment: Accepted to CVPR2019 10 pages, 7 figures

This paper discusses several optical elements now in use at BL43LXU, the RIKEN Quantum NanoDynamics Beamline, of the RIKEN SPring-8 Center. BL43LXU is dedicated to meV-resolved inelastic x-rays scattering using spherical analyzers operating between 15.8 and 25.7 keV. The work described here is relevant for setups on the high-resolution spectrometer (10m two-theta arm) with resolution between 2.8 and 0.8 meV. Specific optics discussed include a multilayer Kirkpatrick-Baez (KB) mirror pair that focuses the full (~1x3mm2) beam at 17.79 keV to a 4.4 X 4.1 {\mu}m2 spot with ~60% throughput, two different types of Soller slits that help reduce backgrounds, masks for the analyzers that allow increased solid angle to be collected while preserving momentum resolution, and a diamond quadrant beam position monitor (BPM). These elements have been used and facilitate experiments in extreme conditions with diamond anvil cells, and liquid measurements at low momentum transfers, among other work., Comment: 7 Pages, 6 figures, for the proceedings of SRI2018 (v2 = minor changes for consistency) This revision corrects some typos and dimensions of Soller Screen

The aim of this study is to clarify the physics which governs the transition from epsilon phase to zeta phase of solid oxygen observed experimentally at 96 GPa using density functional theory (DFT). The transition was predicted at 40 GPa with PBE functional. Then the Hubbard correction was added to enhance the localization of p-orbital of oxygen. The epsilon-zeta transition pressure was significantly improved to 70 GPa. Finally, we included the non-local van der Waals correction. The transition pressure slightly increases to 80 GPa. These results demonstrate that the contribution from Hubbard term is superior to van der Waals term., Comment: 4 pages, 4 figure

In this paper, we propose a method for cloud removal from visible light RGB satellite images by extending the conditional Generative Adversarial Networks (cGANs) from RGB images to multispectral images. Satellite images have been widely utilized for various purposes, such as natural environment monitoring (pollution, forest or rivers), transportation improvement and prompt emergency response to disasters. However, the obscurity caused by clouds makes it unstable to monitor the situation on the ground with the visible light camera. Images captured by a longer wavelength are introduced to reduce the effects of clouds. Synthetic Aperture Radar (SAR) is such an example that improves visibility even the clouds exist. On the other hand, the spatial resolution decreases as the wavelength increases. Furthermore, the images captured by long wavelengths differs considerably from those captured by visible light in terms of their appearance. Therefore, we propose a network that can remove clouds and generate visible light images from the multispectral images taken as inputs. This is achieved by extending the input channels of cGANs to be compatible with multispectral images. The networks are trained to output images that are close to the ground truth using the images synthesized with clouds over the ground truth as inputs. In the available dataset, the proportion of images of the forest or the sea is very high, which will introduce bias in the training dataset if uniformly sampled from the original dataset. Thus, we utilize the t-Distributed Stochastic Neighbor Embedding (t-SNE) to improve the problem of bias in the training dataset. Finally, we confirm the feasibility of the proposed network on the dataset of four bands images, which include three visible light bands and one near-infrared (NIR) band.