Yu Hu

Yu Hu

Auto-generated profile, based on publicly available information

Yu Hu

Institute of Computing Technology Chinese Academy of Sciences

Research Focus & Metrics

Academic Career Length

23Years (2004-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
4 July 2018 in IEEE Robotics and Automation Letters
Verified
Journal Article
Closed access
YZYiming Zeng
Yiming Zeng
YHYu Hu
Yu Hu
SLShice Liu
Shice Liu
JYJing Ye
Jing Ye
YHYinhe Han
Yinhe Han
NSNinghui Sun
Ninghui Sun
For autonomous driving, vehicle detection is the prerequisite for many tasks like collision avoidance and path planning. In this letter, we present a real-time three-dimensional (RT3D) vehicle detection method that utilizes pure LiDAR point cloud to predict the location, orientation, and size of vehicles. In contrast to previous 3-D object detection methods, we used a pre-RoIpooling convolution technique that moves a majority of the convolution operations to ahead of the RoI pooling, leaving just a small part behind, so that significantly boosts the computation efficiency. We also propose a pose-sensitive feature map design which can be strongly activated by the relative poses of vehicles, leading to a high regression accuracy on the location, orientation, and size of vehicles. Experiments on the KITTI benchmark dataset show that the RT3D is not only able to deliver competitive detection accuracy against state-of-the-art methods, but also the first LiDAR-based 3-D vehicle detection work that completes detection within 0.09 s which is even shorter than the scan period of mainstream LiDAR sensors.

5 June 2025 in Pattern Recognition
Verified
Journal Article
Closed access
FMFangyuan Mao
Fangyuan Mao
JMJilin Mei
Jilin Mei
SLShun Lu
Shun Lu
FLFuyang Liu
Fuyang Liu
LCLiang Chen
Liang Chen
YHYu Hu
Yu Hu

1 October 2018
Verified
Conference Paper
Closed access
BJBeibei Jin
Beibei Jin
YHYu Hu
Yu Hu
YZYiming Zeng
Yiming Zeng
QTQiankun Tang
Qiankun Tang
SLShice Liu
Shice Liu
JYJing Ye
Jing Ye
Unsupervised video prediction is a very challenging task due to the complexity and diversity in natural scenes. Prior works directly predicting pixels or optical flows either have the blurring problem or require additional assumptions. We highlight that the crux for video frame prediction lies in precisely capturing the inter-frame variations which encompass the movement of objects and the evolution of the surrounding environment. We then present an unsupervised video prediction framework - Variation Network (VarNet) to directly predict the variations between adjacent frames which are then fused with current frame to generate the future frame. In addition, we propose an adaptively re-weighting mechanism for loss function to offer each pixel a fair weight according to the amplitude of its variation. Extensive experiments for both short-term and long-term video prediction are implemented on two advanced datasets - KTH and KITTI with two evaluating metrics - PSNR and SSIM. For the KTH dataset, the VarNet outperforms the state-of-the-art works up to 11.9% on PSNR and 9.5% on SSIM. As for the KITTI dataset, the performance boosts are up to 55.1% on PSNR and 15.9% on SSIM. Moreover, we verify that the generalization ability of our model excels other state-of-the-art methods by testing on the unseen CalTech Pedestrian dataset after being trained on the KITTI dataset. Source code and video are available at https://github.com/jinbeibei/VarNet.

24 February 2021 in IEEE Access
Verified
Journal Article
Open access
NSNajmeh Samadiani
Najmeh Samadiani
GHGuangyan Huang
Guangyan Huang
YHYu Hu
Yu Hu
XLXiaowei Li
Xiaowei Li
Facial expressions have been proven to be the most effective way for the brain to recognize human emotions in a variety of contexts. With the exponentially increasing research for emotion detection in recent years, facial expression recognition has become an attractive, hot research topic to identify various basic emotions. Happy emotion is one of such basic emotions with many applications, which is more likely recognized by facial expressions than other emotion measurement instruments (e.g., audio/speech, textual and physiological sensing). Nowadays, most methods have been developed for identifying multiple types of emotions, which aim to achieve the best overall precision for all emotions; it is hard for them to optimize the recognition accuracy for single emotion (e.g., happiness). Only a few methods are designed to recognize single happy emotion captured in the unconstrained videos; however, their limitations lie in that the processing of severe head pose variations has not been considered, and the accuracy is still not satisfied. In this paper, we propose a Happy Emotion Recognition model using the 3D hybrid deep and distance features (HappyER-DDF) method to improve the accuracy by utilizing and extracting two different types of deep visual features. First, we employ a hybrid 3D Inception-ResNet neural network and long-short term memory (LSTM) to extract dynamic spatial-temporal features among sequential frames. Second, we detect facial landmarks' features and calculate the distance between each facial landmark and a reference point on the face (e.g., nose peak) to capture their changes when a person starts to smile (or laugh). We implement the experiments using both feature-level and decision-level fusion techniques on three unconstrained video datasets. The results demonstrate that our HappyER-DDF method is arguably more accurate than several currently available facial expression models.

26 June 2023 in Proceedings of the AAAI Conference on Artificial Intelligence
Verified
Journal Article
Open access
PDF file
SLShun Lu
Shun Lu
YHYu Hu
Yu Hu
PWPeihao Wang
Peihao Wang
YHYan Han
Yan Han
JTJianchao Tan
Jianchao Tan
JLJi Liu
Ji Liu
Time-consuming performance evaluation is the bottleneck of traditional Neural Architecture Search (NAS) methods. Predictor-based NAS can speed up performance evaluation by directly predicting performance, rather than training a large number of sub-models and then validating their performance. Most predictor-based NAS approaches use a proxy dataset to train model-based predictors efficiently but suffer from performance degradation and generalization problems. We attribute these problems to the poor abilities of existing predictors to character the sub-models' structure, specifically the topology information extraction and the node feature representation of the input graph data. To address these problems, we propose a Transformer-like NAS predictor PINAT, consisting of a Permutation INvariance Augmentation module serving as both token embedding layer and self-attention head, as well as a Laplacian matrix to be the positional encoding. Our design produces more representative features of the encoded architecture and outperforms state-of-the-art NAS predictors on six search spaces: NAS-Bench-101, NAS-Bench-201, DARTS, ProxylessNAS, PPI, and ModelNet. The code is available at https://github.com/ShunLu91/PINAT.

29 May 2007 in IEEE Transactions on Very Large Scale Integration (VLSI) Systems
Verified
Journal Article
Closed access
YHYinhe Han
Yinhe Han
YHYu Hu
Yu Hu
XLXiaowei Li
Xiaowei Li
HLHuawei Li
Huawei Li
Anshuman ChandraAnshuman Chandra
Anshuman Chandra
An embedded test stimulus decompressor is presented for the test patterns decompression, which can reduce the required channels and vector memory of automatic test equipment (ATE) for complex processor circuit. The proposed decompressor mainly consists of a periodically alterable MUX network which has multiple configurations to decode the input information flexibly and efficiently. In order to reduce the number of test patterns and configurations, a test patterns compaction algorithm, using CI-Graph merging, is proposed. With the proposed periodically alterable MUX network and the patterns compaction algorithm, smaller test data volume and required external pins can be achieved as compared to previous techniques