Bingrui Geng

Ms. Bingrui Geng


Research Focus & Metrics

Academic Career Length

8Years (2019-Present)

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

Representative Publications

View All Publications
1 February 2019 in Information Sciences
Verified
Journal Article
Closed access
BGBingrui Geng
Bingrui Geng
LJLicheng Jiao
Licheng Jiao
MGMaoguo Gong
Maoguo Gong
LLLingling Li
Lingling Li
Yue WuYue Wu
Yue Wu

22 January 2020 in IEEE Journal of Selected Topics in Signal Processing
Verified
Journal Article
Closed access
Yiping DuanYiping Duan
Yiping Duan
CHChaoyi Han
Chaoyi Han
XTXiaoming Tao
Xiaoming Tao
BGBingrui Geng
Bingrui Geng
YDYunfei Du
Yunfei Du
JLJianhua Lu
Jianhua Lu
The 360-degree video/image, also called an omnidirectional video/image or panoramic video/image, is very important in some emerging areas such as virtual reality (VR). Therefore, corresponding image generation algorithms are urgently needed. However, existing image generation models mainly focus on 2-D images and do not consider the spherical structures of panoramic images. In this paper, we propose a panoramic image generation method based on spherical convolution and generative adversarial networks, called spherical generative adversarial networks (SGANs). We adopt the sketch map as the input, which is a concise geometric structure representation of the panoramic image, e.g., comprising approximately 7% of the pixels for a 583×1163 image. Through adversarial learning, a realistic-looking, plausible and high-fidelity spherical image can be obtained from the sparse sketch map. In particular, we build a dataset of the sketch maps using a visual computation-based sketching model. Then, by optimizing SGANs with GAN loss, feature matching loss and perceptual loss, realistic textures and details are recovered gradually. On one hand, it is an improvement using the sparse sketch map as input rather than the denser input, e.g., the features of the textures and colors. On the other hand, spherical convolution helps to remedy space-varying distortions of the planar projection. We conduct extensive experiments on some public panoramic image datasets and compare them with state-of-the-art techniques to validate the superior performance of the proposed approach.

24 January 2026 in Pattern Recognition
Verified
Journal Article
Closed access
SHShuzhan Hu
Shuzhan Hu
MLMingyu Li
Mingyu Li
YLYang Liu
Yang Liu
Weiwei JiangWeiwei Jiang
Weiwei Jiang
BGBingrui Geng
Bingrui Geng
LYLong Ye
Long Ye

12 January 2026 in IEEE Transactions on Affective Computing
Verified
Journal Article
Closed access
BGBingrui Geng
Bingrui Geng
QLQiao Lan
Qiao Lan
CWCheng Wang
Cheng Wang
MWMengyuan Wei
Mengyuan Wei
Current electroencephalogram (EEG)-based emotion recognition research confronts two main challenges. On the one hand, most EEG emotion datasets have limited stimulus diversity and coarse-grained emotional labeling. On the other hand, there is a contradiction between increasingly complex network models and the limited resources of EEG training data. To address these issues, this paper builds a fine-grained EEG emotion dataset for Chinese storytelling (FEEDC) and proposes the lite residual convolution-based transformer neural network (Lite-Resformer), which deeply explores feature information in the time-frequency-space domain of EEG data to improve emotion recognition accuracy, especially positive emotion. First, a fine-grained EEG emotion dataset is constructed using videos from mainstream Chinese social media platforms, covering emotional stimuli related to new productive forces, natural landscapes, and traditional culture. Valid EEG signals and subjective self-assessment data from 65 participants are recorded. Next, a 3D brain map feature superposition method is used to effectively fuse and characterize the frequency, space, and time information of EEG signals. Despite the small amount of EEG data, its rich emotional information motivates a lightweight residual network and Transformer-based network named the Lite-Resformer model, which effectively combines electrode location, spatial, and temporal information from EEG signals, extracting both local and global features. The proposed model has been validated for its effectiveness in valence-arousal emotion classification tasks and achieves an accuracy of 85.48% in the eight-class classification task for discrete emotions, outperforming the existing state-of-the-art methods. This paper provides an effective solution for lightweight finegrained EEG emotion recognition and is of significant value for exploring the emotional factors influencing the spread of Chinese stories