Lane detection plays a key role in autonomous driving. While car cameras always take streaming videos on the way, current lane detection works mainly focus on individual images (frames) by ignoring dynamics along the video. In this work, we collect a
new video instance lane detection (VIL-100) dataset, which contains 100 videos with in total 10,000 frames, acquired from different real traffic scenarios. All the frames in each video are manually annotated to a high-quality instance-level lane annotation, and a set of frame-level and video-level metrics are included for quantitative performance evaluation. Moreover, we propose a new baseline model, named multi-level memory aggregation network (MMA-Net), for video instance lane detection. In our approach, the representation of current frame is enhanced by attentively aggregating both local and global memory features from other frames. Experiments on the new collected dataset show that the proposed MMA-Net outperforms state-of-the-art lane detection methods and video object segmentation methods. We release our dataset and code at https://github.com/yujun0-0/MMA-Net.
Retinal vessel segmentation plays a key role in computer-aided screening, diagnosis, and treatment of various cardiovascular and ophthalmic diseases. Recently, deep learning-based retinal vessel segmentation algorithms have achieved remarkable perfor
mance. However, due to the domain shift problem, the performance of these algorithms often degrades when they are applied to new data that is different from the training data. Manually labeling new data for each test domain is often a time-consuming and laborious task. In this work, we explore unsupervised domain adaptation in retinal vessel segmentation by using entropy-based adversarial learning and transfer normalization layer to train a segmentation network, which generalizes well across domains and requires no annotation of the target domain. Specifically, first, an entropy-based adversarial learning strategy is developed to reduce the distribution discrepancy between the source and target domains while also achieving the objective of entropy minimization on the target domain. In addition, a new transfer normalization layer is proposed to further boost the transferability of the deep network. It normalizes the features of each domain separately to compensate for the domain distribution gap. Besides, it also adaptively selects those feature channels that are more transferable between domains, thus further enhancing the generalization performance of the network. We conducted extensive experiments on three regular fundus image datasets and an ultra-widefield fundus image dataset, and the results show that our approach yields significant performance gains compared to other state-of-the-art methods.
The Fourier transform spectrometer (FTS) is a core instrument for solar observation with high spectral resolution, especially in the infrared. The Infrared System for the Accurate Measurement of Solar Magnetic Field (AIMS), working at 10-13 $mu m$, w
ill use a FTS to observe the solar spectrum. The Bruker IFS-125HR, which meets the spectral resolution requirement of AIMS but just equips with a point source detector, is employed to carry out preliminary experiment for AIMS. A sun-light feeding experimental system is further developed. Several experiments are taken with them during 2018 and 2019 to observe the solar spectrum in the visible and near infrared wavelength, respectively. We also proposed an inversion method to retrieve the solar spectrum from the observed interferogram and compared it with the standard solar spectrum atlas. Although there is a wavelength limitation due to the present sun-light feeding system, the results in the wavelength band from 0.45-1.0 $mu m$ and 1.0-2.2 $mu m$ show a good consistence with the solar spectrum atlas, indicating the validity of our observing configuration, the data analysis method and the potential to work in longer wavelength. The work provided valuable experience for the AIMS not only for the operation of a FTS but also for the development of its scientific data processing software.
The celebrated Seq2Seq technique and its numerous variants achieve excellent performance on many tasks such as neural machine translation, semantic parsing, and math word problem solving. However, these models either only consider input objects as se
quences while ignoring the important structural information for encoding, or they simply treat output objects as sequence outputs instead of structural objects for decoding. In this paper, we present a novel Graph-to-Tree Neural Networks, namely Graph2Tree consisting of a graph encoder and a hierarchical tree decoder, that encodes an augmented graph-structured input and decodes a tree-structured output. In particular, we investigated our model for solving two problems, neural semantic parsing and math word problem. Our extensive experiments demonstrate that our Graph2Tree model outperforms or matches the performance of other state-of-the-art models on these tasks.
Superconducting qubits provide a competitive platform for quantum simulation of complex dynamics that lies at the heart of quantum many-body systems, because of the flexibility and scalability afforded by the nature of microfabrication. However, in a
multiqubit device, the physical form of couplings between qubits is either an electric (capacitor) or magnetic field (inductor), and the associated quadratic field energy determines that only two-body interaction in the Hamiltonian can be directly realized. Here we propose and experimentally synthesize the three-body spin-chirality interaction in a superconducting circuit based on Floquet engineering. By periodically modulating the resonant frequencies of the qubits connected with each other via capacitors, we can dynamically turn on and off qubit-qubit couplings, and further create chiral flows of the excitations in the three-qubit circular loop. Our result is a step toward engineering dynamical and many-body interactions in multiqubit superconducting devices, which potentially expands the degree of freedom in quantum simulation tasks.
Superradiance and subradiance concerning enhanced and inhibited collective radiation of an ensemble of atoms have been a central topic in quantum optics. However, precise generation and control of these states remain challenging. Here we deterministi
cally generate up to 10-qubit superradiant and 8-qubit subradiant states, each containing a single excitation, in a superconducting quantum circuit with multiple qubits interconnected by a cavity resonator. The $sqrt{N}$-scaling enhancement of the coupling strength between the superradiant states and the cavity is validated. By applying appropriate phase gate on each qubit, we are able to switch the single collective excitation between superradiant and subradiant states. While the subradiant states containing a single excitation are forbidden from emitting photons, we demonstrate that they can still absorb photons from the resonator. However, for even number of qubits, a singlet state with half of the qubits being excited can neither emit nor absorb photons, which is verified with 4 qubits. This study is a step forward in coherent control of collective radiation and has promising applications in quantum information processing.
Human-object interactions (HOI) recognition and pose estimation are two closely related tasks. Human pose is an essential cue for recognizing actions and localizing the interacted objects. Meanwhile, human action and their interacted objects localiza
tions provide guidance for pose estimation. In this paper, we propose a turbo learning framework to perform HOI recognition and pose estimation simultaneously. First, two modules are designed to enforce message passing between the tasks, i.e. pose aware HOI recognition module and HOI guided pose estimation module. Then, these two modules form a closed loop to utilize the complementary information iteratively, which can be trained in an end-to-end manner. The proposed method achieves the state-of-the-art performance on two public benchmarks including Verbs in COCO (V-COCO) and HICO-DET datasets.
B-mode ultrasound tongue imaging is widely used in the speech production field. However, efficient interpretation is in a great need for the tongue image sequences. Inspired by the recent success of unsupervised deep learning approach, we explore uns
upervised convolutional network architecture for the feature extraction in the ultrasound tongue image, which can be helpful for the clinical linguist and phonetics. By quantitative comparison between different unsupervised feature extraction approaches, the denoising convolutional autoencoder (DCAE)-based method outperforms the other feature extraction methods on the reconstruction task and the 2010 silent speech interface challenge. A Word Error Rate of 6.17% is obtained with DCAE, compared to the state-of-the-art value of 6.45% using Discrete cosine transform as the feature extractor. Our codes are available at https://github.com/DeePBluE666/Source-code1.
This paper proposes a graph computation based sequential power flow calculation method for Line Commutated Converter (LCC) based large-scale AC/DC systems to achieve a high computing performance. Based on the graph theory, the complex AC/DC system is
first converted to a graph model and stored in a graph database. Then, the hybrid system is divided into several isolated areas with graph partition algorithm by decoupling AC and DC networks. Thus, the power flow analysis can be executed in parallel for each independent area with the new selected slack buses. Furthermore, for each area, the node-based parallel computing (NPC) and hierarchical parallel computing (HPC) used in graph computation are employed to speed up fast decoupled power flow (FDPF). Comprehensive case studies on the IEEE 300-bus, polished South Carolina 12,000-bus system and a China 11,119-bus system are performed to demonstrate the accuracy and efficiency of the proposed method
Power flow analysis plays a fundamental and critical role in the energy management system (EMS). It is required to well accommodate large and complex power system. To achieve a high performance and accurate power flow analysis, a graph computing base
d distributed power flow analysis approach is proposed in this paper. Firstly, a power system network is divided into multiple areas. Slack buses are selected for each area and, at each SCADA sampling period, the inter-area transmission line power flows are equivalently allocated as extra load injections to corresponding buses. Then, the system network is converted into multiple independent areas. In this way, the power flow analysis could be conducted in parallel for each area and the solved system states could be guaranteed without compromise of accuracy. Besides, for each area, graph computing based fast decoupled power flow (FDPF) is employed to quickly analyze system states. IEEE 118-bus system and MP 10790-bus system are employed to verify the results accuracy and present the promising computation performance of the proposed approach.
