Jinlong Lin
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Jinlong Lin has a total of 24 co-authors. Here are the top co-authors by number of shared publications:
- Boya Di
Peking University
2 shared publications - Lingyang Song
Peking University
2 shared publications - Xin Li
Nanyang Technological University
2 shared publications - Yutong Zhang
Future Research Laboratory, China Mobile Research Institute, Beijing, China
2 shared publications Changjin Li
Peking University
1 shared publication- Chao Zhang
Peking University
1 shared publication
Latest Publications
- 26 November 2023
- Published by Springer Nature
- p. 75-85
Design of a Multimodal Short Video Classification Model
- 26 November 2023
- Published by Springer Nature
- p. 75-85
Design of a Multimodal Short Video Classification Model
- 1 July 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Affective Computing
- Vol. 14 (2) , 1391-1403
Multimodal Affective States Recognition Based on Multiscale CNNs and Biologically Inspired Decision Fusion Model
There has been an encouraging progress in the affective states recognition models based on the single-modality signals as electroencephalogram (EEG) signals or peripheral physiological signals in recent years. However, multimodal physiological signals-based affective states recognition methods have not been thoroughly exploited yet. Here we propose Multiscale Convolutional Neural Networks (Multiscale CNNs) and a biologically inspired decision fusion model for multimodal affective states recognition. Firstly, the raw signals are pre-processed with baseline signals. Then, the High Scale CNN and Low Scale CNN in Multiscale CNNs are utilized to predict the probability of affective states output for EEG and each peripheral physiological signal respectively. Finally, the fusion model calculates the reliability of each single-modality signals by the Euclidean distance between various class labels and the classification probability from Multiscale CNNs, and the decision is made by the more reliable modality information while other modalities information is retained. We use this model to classify four affective states from the arousal valence plane in the DEAP and AMIGOS dataset. The results show that the fusion model improves the accuracy of affective states recognition significantly compared with the result on single-modality signals, and the recognition accuracy of the fusion result achieve 98.52% and 99.89% in the DEAP and AMIGOS dataset respectively.
- 1 July 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Affective Computing
- Vol. 14 (2) , 1391-1403
Multimodal Affective States Recognition Based on Multiscale CNNs and Biologically Inspired Decision Fusion Model
There has been an encouraging progress in the affective states recognition models based on the single-modality signals as electroencephalogram (EEG) signals or peripheral physiological signals in recent years. However, multimodal physiological signals-based affective states recognition methods have not been thoroughly exploited yet. Here we propose Multiscale Convolutional Neural Networks (Multiscale CNNs) and a biologically inspired decision fusion model for multimodal affective states recognition. Firstly, the raw signals are pre-processed with baseline signals. Then, the High Scale CNN and Low Scale CNN in Multiscale CNNs are utilized to predict the probability of affective states output for EEG and each peripheral physiological signal respectively. Finally, the fusion model calculates the reliability of each single-modality signals by the Euclidean distance between various class labels and the classification probability from Multiscale CNNs, and the decision is made by the more reliable modality information while other modalities information is retained. We use this model to classify four affective states from the arousal valence plane in the DEAP and AMIGOS dataset. The results show that the fusion model improves the accuracy of affective states recognition significantly compared with the result on single-modality signals, and the recognition accuracy of the fusion result achieve 98.52% and 99.89% in the DEAP and AMIGOS dataset respectively.
- Open Access
Leveraging Planar Regularities for Point Line Visual-Inertial Odometry
With monocular Visual-Inertial Odometry (VIO) system, 3D point cloud and camera motion can be estimated simultaneously. Because pure sparse 3D points provide a structureless representation of the environment, generating 3D mesh from sparse points can further model the environment topology and produce dense mapping. To improve the accuracy of 3D mesh generation and localization, we propose a tightly-coupled monocular VIO system, PLP-VIO, which exploits point features and line features as well as plane regularities. The co-planarity constraints are used to leverage additional structure information for the more accurate estimation of 3D points and spatial lines in state estimator. To detect plane and 3D mesh robustly, we combine both the line features with point features in the detection method. The effectiveness of the proposed method is verified on both synthetic data and public datasets and is compared with other state-of-the-art algorithms.
Open AccessLeveraging Planar Regularities for Point Line Visual-Inertial Odometry
With monocular Visual-Inertial Odometry (VIO) system, 3D point cloud and camera motion can be estimated simultaneously. Because pure sparse 3D points provide a structureless representation of the environment, generating 3D mesh from sparse points can further model the environment topology and produce dense mapping. To improve the accuracy of 3D mesh generation and localization, we propose a tightly-coupled monocular VIO system, PLP-VIO, which exploits point features and line features as well as plane regularities. The co-planarity constraints are used to leverage additional structure information for the more accurate estimation of 3D points and spatial lines in state estimator. To detect plane and 3D mesh robustly, we combine both the line features with point features in the detection method. The effectiveness of the proposed method is verified on both synthetic data and public datasets and is compared with other state-of-the-art algorithms.
- Open Access
Beyond Cell-free MIMO: Energy Efficient Reconfigurable Intelligent Surface Aided Cell-free MIMO Communications
Cell-free systems can effectively eliminate the inter-cell interference by enabling multiple base stations (BSs) to cooperatively serve users without cell boundaries at the expense of high costs of hardware and power sources due to the large-scale deployment of BSs. To tackle this issue, the low-cost reconfigurable intelligent surface (RIS) can serve as a promising technique to improve the energy efficiency of cell-free systems. In this paper, we consider an RIS aided cell-free MIMO system where multiple RISs are deployed around BSs and users to create favorable propagation conditions via reconfigurable reflections in a low-cost way, thereby enhancing cell-free MIMO communications. To maximize the energy efficiency, a hybrid beamforming (HBF) scheme consisting of the digital beamforming at BSs and the RIS-based analog beamforming is proposed. The energy efficiency maximization problem is formulated and an iterative algorithm is designed to solve this problem. The impact of the transmit power, the number of RIS, and the RIS size on energy efficiency are investigated. Both theoretical analysis and simulation results reveal that the optimal energy efficiency depends on the numbers of RISs and the RIS size. Numerical evaluations also show that the proposed system can achieve a higher energy efficiency than conventional ones.
Open AccessBeyond Cell-free MIMO: Energy Efficient Reconfigurable Intelligent Surface Aided Cell-free MIMO Communications
Cell-free systems can effectively eliminate the inter-cell interference by enabling multiple base stations (BSs) to cooperatively serve users without cell boundaries at the expense of high costs of hardware and power sources due to the large-scale deployment of BSs. To tackle this issue, the low-cost reconfigurable intelligent surface (RIS) can serve as a promising technique to improve the energy efficiency of cell-free systems. In this paper, we consider an RIS aided cell-free MIMO system where multiple RISs are deployed around BSs and users to create favorable propagation conditions via reconfigurable reflections in a low-cost way, thereby enhancing cell-free MIMO communications. To maximize the energy efficiency, a hybrid beamforming (HBF) scheme consisting of the digital beamforming at BSs and the RIS-based analog beamforming is proposed. The energy efficiency maximization problem is formulated and an iterative algorithm is designed to solve this problem. The impact of the transmit power, the number of RIS, and the RIS size on energy efficiency are investigated. Both theoretical analysis and simulation results reveal that the optimal energy efficiency depends on the numbers of RISs and the RIS size. Numerical evaluations also show that the proposed system can achieve a higher energy efficiency than conventional ones.
- 28 September 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 5 (4) , 6972-6979
Co-Planar Parametrization for Stereo-SLAM and Visual-Inertial Odometry
This work proposes a novel SLAM framework for stereo and visual inertial odometry estimation. It builds an efficient and robust parametrization of co-planar points and lines which leverages specific geometric constraints to improve camera pose optimization in terms of both efficiency and accuracy. The pipeline consists of extracting 2D points and lines, predicting planar regions and filtering the outliers via RANSAC. Our parametrization scheme then represents co-planar points and lines as their 2D image coordinates and parameters of planes. We demonstrate the effectiveness of the proposed method by comparing it to traditional parametrizations in a novel Monte-Carlo simulation set. Further, the whole stereo SLAM and VIO system is compared with state-of-the-art methods on the public real-world dataset EuRoC. Our method shows better results in terms of accuracy and efficiency than the state-of-the-art. We plan to publicly release the code of our SLAM framework.
- 28 September 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 5 (4) , 6972-6979
Co-Planar Parametrization for Stereo-SLAM and Visual-Inertial Odometry
This work proposes a novel SLAM framework for stereo and visual inertial odometry estimation. It builds an efficient and robust parametrization of co-planar points and lines which leverages specific geometric constraints to improve camera pose optimization in terms of both efficiency and accuracy. The pipeline consists of extracting 2D points and lines, predicting planar regions and filtering the outliers via RANSAC. Our parametrization scheme then represents co-planar points and lines as their 2D image coordinates and parameters of planes. We demonstrate the effectiveness of the proposed method by comparing it to traditional parametrizations in a novel Monte-Carlo simulation set. Further, the whole stereo SLAM and VIO system is compared with state-of-the-art methods on the public real-world dataset EuRoC. Our method shows better results in terms of accuracy and efficiency than the state-of-the-art. We plan to publicly release the code of our SLAM framework.
- 21 February 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Vehicular Technology
- Vol. 69 (4) , 4388-4400
HetMEC: Heterogeneous Multi-Layer Mobile Edge Computing in the 6 G Era
Driven by an increasing number of mobile applications, mobile edge computing (MEC) has been considered as a promising candidate to support the huge amount of data processing services. However, the conventional MEC suffers from the insufficient utilization of computing and transmission resources through the entire network, resulting in inevitably long processing and transmission latency especially to computation-intensive applications in the 6G era. In this article, we propose a heterogeneous multi-layer mobile edge computing (HetMEC), where different devices, ranging from edge devices (EDs), i.e., the mobile devices that generate raw data of computing tasks in the radio access networks, to the cloud center (CC), are inherently involved different layers of the network and collaborate for data processing. To support a low-latency service, a reinforcement learning-based framework is constructed to adapt to the unstable wireless environments as well as the dynamically varying data generation speed of each ED. Under this framework, key research issues and solutions including task offloading, cognitive radio based spectrum access, pricing scheme design, and network congestion control are presented. Some further research directions and opening issues are also discussed in the perception of network planning and optimization, network control, and application-specific issues.
- 21 February 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Vehicular Technology
- Vol. 69 (4) , 4388-4400
HetMEC: Heterogeneous Multi-Layer Mobile Edge Computing in the 6 G Era
Driven by an increasing number of mobile applications, mobile edge computing (MEC) has been considered as a promising candidate to support the huge amount of data processing services. However, the conventional MEC suffers from the insufficient utilization of computing and transmission resources through the entire network, resulting in inevitably long processing and transmission latency especially to computation-intensive applications in the 6G era. In this article, we propose a heterogeneous multi-layer mobile edge computing (HetMEC), where different devices, ranging from edge devices (EDs), i.e., the mobile devices that generate raw data of computing tasks in the radio access networks, to the cloud center (CC), are inherently involved different layers of the network and collaborate for data processing. To support a low-latency service, a reinforcement learning-based framework is constructed to adapt to the unstable wireless environments as well as the dynamically varying data generation speed of each ED. Under this framework, key research issues and solutions including task offloading, cognitive radio based spectrum access, pricing scheme design, and network congestion control are presented. Some further research directions and opening issues are also discussed in the perception of network planning and optimization, network control, and application-specific issues.
- 1 June 2007
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- No. 0747668X,p. 1-5
Recover Chromatic Information of Images Spoiled by Near Infrared Radiation
CMOS imaging devices are widely used in digital image and video products. However, the strong responses to near infrared radiation of device's R, G and B channels cause the chromatic difference between the captured images and their original scenes. In this paper, we present a new method to solve the problem. After finding the reference pixels with the least value of G or B, we estimated the global ratio of near infrared radiation response to red radiation response. Then, the additive response to near infrared radiation was subtracted from the image's raw data of three channels, and the original scene's chromatic information was recovered for all pixels. The experimental result shows that the proposed algorithms can eliminate the effects of infrared radiation and recover chromatic saturation of original scenes without optical high pass filter.
- 1 June 2007
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- No. 0747668X,p. 1-5
Recover Chromatic Information of Images Spoiled by Near Infrared Radiation
CMOS imaging devices are widely used in digital image and video products. However, the strong responses to near infrared radiation of device's R, G and B channels cause the chromatic difference between the captured images and their original scenes. In this paper, we present a new method to solve the problem. After finding the reference pixels with the least value of G or B, we estimated the global ratio of near infrared radiation response to red radiation response. Then, the additive response to near infrared radiation was subtracted from the image's raw data of three channels, and the original scene's chromatic information was recovered for all pixels. The experimental result shows that the proposed algorithms can eliminate the effects of infrared radiation and recover chromatic saturation of original scenes without optical high pass filter.