Greg Wilsbacher
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Greg Wilsbacher has a total of 8 co-authors. Here are the top co-authors by number of shared publications:
- Song Wang
Faculty of Computer Science and Artificial Intelligence, Shenzhen University of Advanced Technology, Shenzhen, Guangdong, China
3 shared publications - Jun Zhou
Computer Science and Engineering Department, University of South Carolina, Columbia, SC 29425, USA
2 shared publications - Liang Zhao
Computer Science & Engineering Department, University of South Carolina, Columbia, SC 29201, USA
2 shared publications - Lujun Zhai
Department of Electrical and Computer Engineering, Prairie View A&M University, Prairie View, TX 77446, USA
2 shared publications - Suxia Cui
Department of Electrical and Computer Engineering, Prairie View A&M University, Prairie View, TX 77446, USA
2 shared publications - Yonghui Wang
Department of Computer Science, Prairie View A&M University, Prairie View, TX 77446, USA
2 shared publications
Latest Publications
- Open Access
- 14 November 2026
- Published by MDPI AG in Journal of Imaging
- Vol. 12 (11) , 578
HistoSR: An Approach to Historical Imagery Super-Resolution
Historical photographs and films provide valuable visual records of the past, yet their quality is frequently compromised by limited imaging capabilities and complex acquisition-related distortions. Recovering high-resolution details from these materials remains difficult because conventional super-resolution (SR) approaches typically rely on simplified degradation assumptions and paired low-resolution (LR) and high-resolution (HR) training samples, which are generally unavailable for historical imagery. To address these limitations, we develop HistoSR, a two-stage restoration framework that combines unpaired degradation correction with pseudo-paired SR reconstruction. In the first stage, Transformer-based generators are incorporated into a CycleGAN architecture to learn the translation between degraded historical LR images and clean LR images without requiring aligned samples. The second stage learns the LR-to-HR reconstruction mapping using paired training samples derived from modern HR images. This design enables the restoration of historical imagery without relying on corresponding historical HR ground truth. We construct a dataset comprising 20,000 historical face images collected from World War II documentary films to evaluate the proposed approach. On the historical face dataset, HistoSR obtains a Neural Side-by-Side predicted preference score of 0.27 and a NIQE value of 5.43. When trained on DIV2K and evaluated on Set14 at a ×4 scaling factor, the framework achieves 29.78 dB PSNR and 0.8103 SSIM, outperforming the evaluated competing methods. These findings demonstrate the effectiveness of HistoSR for historical face restoration and its potential applicability to general image super-resolution.
- 14 November 2026
- Published by MDPI AG in Journal of Imaging
- Vol. 12 (11) , 578
Open AccessHistoSR: An Approach to Historical Imagery Super-Resolution
Historical photographs and films provide valuable visual records of the past, yet their quality is frequently compromised by limited imaging capabilities and complex acquisition-related distortions. Recovering high-resolution details from these materials remains difficult because conventional super-resolution (SR) approaches typically rely on simplified degradation assumptions and paired low-resolution (LR) and high-resolution (HR) training samples, which are generally unavailable for historical imagery. To address these limitations, we develop HistoSR, a two-stage restoration framework that combines unpaired degradation correction with pseudo-paired SR reconstruction. In the first stage, Transformer-based generators are incorporated into a CycleGAN architecture to learn the translation between degraded historical LR images and clean LR images without requiring aligned samples. The second stage learns the LR-to-HR reconstruction mapping using paired training samples derived from modern HR images. This design enables the restoration of historical imagery without relying on corresponding historical HR ground truth. We construct a dataset comprising 20,000 historical face images collected from World War II documentary films to evaluate the proposed approach. On the historical face dataset, HistoSR obtains a Neural Side-by-Side predicted preference score of 0.27 and a NIQE value of 5.43. When trained on DIV2K and evaluated on Set14 at a ×4 scaling factor, the framework achieves 29.78 dB PSNR and 0.8103 SSIM, outperforming the evaluated competing methods. These findings demonstrate the effectiveness of HistoSR for historical face restoration and its potential applicability to general image super-resolution.
- Open Access
- 20 September 2024
- Published by MDPI AG in Journal of Imaging
- Vol. 10 (9) , 236
Historical Blurry Video-Based Face Recognition
Face recognition is a widely used computer vision, which plays an increasingly important role in user authentication systems, security systems, and consumer electronics. The models for most current applications are based on high-definition digital cameras. In this paper, we focus on digital images derived from historical motion picture films. Historical motion picture films often have poorer resolution than modern digital imagery, making face detection a more challenging task. To approach this problem, we first propose a trunk–branch concatenated multi-task cascaded convolutional neural network (TB-MTCNN), which efficiently extracts facial features from blurry historical films by combining the trunk with branch networks and employing various sizes of kernels to enrich the multi-scale receptive field. Next, we build a deep neural network-integrated object-tracking algorithm to compensate for failed recognition over one or more video frames. The framework combines simple online and real-time tracking with deep data association (Deep SORT), and TB-MTCNN with the residual neural network (ResNet) model. Finally, a state-of-the-art image restoration method is employed to reduce the effect of noise and blurriness. The experimental results show that our proposed joint face recognition and tracking network can significantly reduce missed recognition in historical motion picture film frames.
- 20 September 2024
- Published by MDPI AG in Journal of Imaging
- Vol. 10 (9) , 236
Open AccessHistorical Blurry Video-Based Face Recognition
Face recognition is a widely used computer vision, which plays an increasingly important role in user authentication systems, security systems, and consumer electronics. The models for most current applications are based on high-definition digital cameras. In this paper, we focus on digital images derived from historical motion picture films. Historical motion picture films often have poorer resolution than modern digital imagery, making face detection a more challenging task. To approach this problem, we first propose a trunk–branch concatenated multi-task cascaded convolutional neural network (TB-MTCNN), which efficiently extracts facial features from blurry historical films by combining the trunk with branch networks and employing various sizes of kernels to enrich the multi-scale receptive field. Next, we build a deep neural network-integrated object-tracking algorithm to compensate for failed recognition over one or more video frames. The framework combines simple online and real-time tracking with deep data association (Deep SORT), and TB-MTCNN with the residual neural network (ResNet) model. Finally, a state-of-the-art image restoration method is employed to reduce the effect of noise and blurriness. The experimental results show that our proposed joint face recognition and tracking network can significantly reduce missed recognition in historical motion picture film frames.
- Open Access
- 21 November 2023
- Published by MDPI AG in Applied Sciences
- Vol. 13 (23) , 12540
CommuSpotter: Scene Text Spotting with Multi-Task Communication
Scene text spotting is a challenging multi-task modulation for locating and recognizing texts in complex scenes. Existing end-to-end text spotters generally adopt sequentially decoupled multi-tasks, consisting of text detection and text recognition modules. Although customized modules are designed to connect the tasks closely, there is no interaction among multiple tasks, resulting in compatible information loss for the overall text spotting. Moreover, the independent and sequential modulation is unidirectional, accumulating errors from early to later tasks. In this paper, we propose CommuSpotter, which enhances multi-task communication by explicitly and concurrently sharing compatible information in overall scene text spotting. To address task-specific inconsistencies, we propose a Conversation Mechanism (CM) to extract and exchange expertise in each specific task with others. Specifically, the detection task is rectified by the text recognition task to filter out duplicated results and false positives, while the text recognition task is corrected by the rectified text detection task to replenish missing characters and decrease non-text interruptions. Consequently, the communication compensates for interaction information and breaks the sequential pipeline of error propagation. In addition, we adopt text semantic segmentation in the text recognition task, which reduces the complex design of customized modules and corresponding extra annotations. Compared with state-of-the-art methods, experimental results show that our method achieves competitive results with computation efficiency.
- 21 November 2023
- Published by MDPI AG in Applied Sciences
- Vol. 13 (23) , 12540
Open AccessCommuSpotter: Scene Text Spotting with Multi-Task Communication
Scene text spotting is a challenging multi-task modulation for locating and recognizing texts in complex scenes. Existing end-to-end text spotters generally adopt sequentially decoupled multi-tasks, consisting of text detection and text recognition modules. Although customized modules are designed to connect the tasks closely, there is no interaction among multiple tasks, resulting in compatible information loss for the overall text spotting. Moreover, the independent and sequential modulation is unidirectional, accumulating errors from early to later tasks. In this paper, we propose CommuSpotter, which enhances multi-task communication by explicitly and concurrently sharing compatible information in overall scene text spotting. To address task-specific inconsistencies, we propose a Conversation Mechanism (CM) to extract and exchange expertise in each specific task with others. Specifically, the detection task is rectified by the text recognition task to filter out duplicated results and false positives, while the text recognition task is corrected by the rectified text detection task to replenish missing characters and decrease non-text interruptions. Consequently, the communication compensates for interaction information and breaks the sequential pipeline of error propagation. In addition, we adopt text semantic segmentation in the text recognition task, which reduces the complex design of customized modules and corresponding extra annotations. Compared with state-of-the-art methods, experimental results show that our method achieves competitive results with computation efficiency.
- 20 October 2022
- Published by Springer Nature
- p. 163-182
Background-Insensitive Scene Text Recognition with Text Semantic Segmentation
- 20 October 2022
- Published by Springer Nature
- p. 163-182
Background-Insensitive Scene Text Recognition with Text Semantic Segmentation