Article
Instruments & Instrumentation
Lei Fu, Wen-bin Gu, Yong-bao Ai, Wei Li, Dong Wang
Summary: An adaptive spatial pixel-level feature fusion network is proposed to effectively fuse information from visible and thermal infrared images, improving pedestrian detection performance. Experimental results demonstrate that the method achieves a good balance between detection speed and accuracy on the KAIST dataset.
INFRARED PHYSICS & TECHNOLOGY
(2021)
Article
Physics, Multidisciplinary
Ching-Hsun Tseng, Shin-Jye Lee, Jianan Feng, Shengzhong Mao, Yu-Ping Wu, Jia-Yu Shang, Xiao-Jun Zeng
Summary: This work proposes an efficient and robust backbone, UPANets, which utilizes channel and spatial direction attentions to expand the receptive fields in shallow convolutional layers. Experimental results show that UPANets achieve better performance with fewer resources on CIFAR-{10, 100} than existing state-of-the-art methods.
Review
Behavioral Sciences
H. Bowman, D. J. Collins, A. K. Nayak, D. Cruse
Summary: Predictive-coding is a influential theory in Neuroscience, but its falsifiability has been questioned. We propose that there are patterns of behavioral and neuroimaging data that can challenge predictive-coding. Furthermore, we argue that the precision-weighting in predictive-coding can explain contra (vanilla) predictive patterns, but this does not render predictive-coding unfalsifiable.
NEUROSCIENCE AND BIOBEHAVIORAL REVIEWS
(2023)
Article
Chemistry, Analytical
Longzhen Yu, Jianhua Zhu, Qian Zhao, Zhixian Wang
Summary: This study proposes an efficient defect inspection algorithm accelerated by FPGA, achieving auto-adaptive operation with the help of AI and machine learning. The algorithm is improved with an attention mechanism and data augmentation to enhance accuracy and efficiency. Deployed on a PYNQ-Z2 FPGA board, it meets the requirements of industrial production.
Article
Chemistry, Analytical
Pankaj Bhowmik, Md Jubaer Hossain Pantho, Christophe Bobda
Summary: This paper presents a hardware architecture for smart cameras that utilizes visual attention-oriented computational strategy and hierarchical processing to improve image processing speed and energy efficiency.
Article
Robotics
Tao Xie, Ke Wang, Ruifeng Li, Xinyue Tang, Lijun Zhao
Summary: In this paper, the authors propose PANet, a pixel-level attention network with embedding vector features, for addressing the challenge of 6D pose estimation from a single RGBD image under severe occlusion. PANet utilizes attention mechanism and a novel selection scheme for robust pose estimation, and achieves significant improvements over existing methods according to extensive experimental results.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2022)
Article
Engineering, Multidisciplinary
Zishuo Dong, Hang Zhang, Allen A. Zhang, Yang Liu, Zhihao Lin, Anzheng He, Changfa Ai
Summary: This paper proposes a robust semantic segmentation algorithm named Marking-DNet for pixel-level recognition of pavement markings. It improves the encoder-decoder architecture of DeepLabV3+ by importing feature maps from four different scales, resulting in enhanced information exchange. Marking-DNet employs the Object-Contextual Representation and the Convolutional Block Attention Module to conduct contextual learning more efficiently and explicitly implement spatial and channel attention. Experimental results demonstrate that Marking-DNet outperforms six state-of-the-art semantic segmentation models in detecting pavement markings using both private and public image datasets, accurately detecting various complex pavement markings except for heavily-worn ones.
Article
Engineering, Electrical & Electronic
Pingcheng Dong, Zhuoyu Chen, Zhuoao Li, Yuzhe Fu, Lei Chen, Fengwei An
Summary: This paper proposes a hardware-oriented SGM algorithm with pixel-level pipeline and region-optimized cost aggregation for high-speed processing and low hardware-resource usage. The algorithm is demonstrated on low-cost XILINX Spartan-7 and advanced Stratix-V FPGA devices for VGA depth estimation, achieving high processing speeds and energy efficiency.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS I-REGULAR PAPERS
(2022)
Article
Horticulture
Zheng Li, Weijie Tao, Jianlei Liu, Fenghua Zhu, Guangyue Du, Guanggang Ji
Summary: In this study, a new self-attention mechanism (GD-Attention) is proposed to improve the focus on disease areas in tomato leaf disease image classification models, achieving competitive performance.
Article
Chemistry, Analytical
Tung M. Luu, Thang Vu, Thanh Nguyen, Chang D. Yoo
Summary: In this study, a visual pretraining via contrastive predictive model (VPCPM) framework is proposed to overcome the limitations of reward-driven representation learning in vision-based reinforcement learning (RL). By training the convolutional encoder with the supervision of contrastive loss, better representations are learned by perceiving the underlying dynamics through a pair of forward and inverse models. Experimental results show that by initializing the encoders with VPCPM, the performance of state-of-the-art vision-based RL algorithms is significantly improved, surpassing or matching the performance of prior unsupervised methods. The learned representations also generalize successfully to new tasks with similar observation and action spaces.
Article
Materials Science, Textiles
Zebin Su, Hao Zhang, Pengfei Li, Huanhuan Zhang, Yanjun Lu
Summary: A lightweight model of digital printing fabric defect detection based on YOLOX is proposed in this paper. By introducing the SE attention module to enhance features, it improves the accuracy of diversified defect detection and solves the influence of small target detection accuracy.
JOURNAL OF ENGINEERED FIBERS AND FABRICS
(2023)
Article
Plant Sciences
Guoqiang Li, Yuchao Wang, Qing Zhao, Peiyan Yuan, Baofang Chang
Summary: This paper proposed an efficient deep learning architecture PMVT based on MobileViT for real-time detection of plant diseases, achieving high accuracy and low cost. Extensive experiments on agricultural datasets demonstrate its superiority in accuracy and performance.
FRONTIERS IN PLANT SCIENCE
(2023)
Article
Chemistry, Analytical
Gui Yu, Juming Dong, Yihang Wang, Xinglin Zhou
Summary: In this paper, we proposed a U-shaped encoder-decoder semantic segmentation network, called RUC-Net, combining Unet and Resnet for pixel-level pavement crack image segmentation, to address the challenges of complex backgrounds, uneven illumination, irregular patterns, and various types of noise interference in crack detection. We introduced the spatial-channel squeeze and excitation (scSE) attention module and the focal loss function to improve the detection effect and handle the class imbalance problem. Evaluation on three public datasets, CFD, Crack500, and DeepCrack, showed that our methods achieved better results compared to FCN, Unet, and SegNet. Additionally, we conducted ablation studies on the CFD dataset, comparing the differences of various scSE modules and their combinations in improving crack detection performance.
Article
Agriculture, Dairy & Animal Science
Jinye Hao, Hongming Zhang, Yamin Han, Jie Wu, Lixiang Zhou, Zhibo Luo, Yutong Du
Summary: Accurate farming is crucial for pasture management and productivity improvement. Sheep face detection based on lightweight convolutional neural network is proposed in this study, showing real-time and robust detection. An improved RetinaFace algorithm is used to achieve accurate and real-time detection of sheep faces on actual sheep farms, demonstrating superior performance.
Article
Computer Science, Information Systems
Parag Bhuyan, Pranav Kumar Singh, Sujit Kumar Das
Summary: Early detection of tea leaf diseases is crucial for maintaining crop yield and agricultural production. This study proposes a deep convolutional neural network model specifically designed for tea leaf disease diagnosis, achieving improved accuracy. Experimental results demonstrate that the proposed model outperforms other standard CNN models and surpasses some recent works in this field.
MULTIMEDIA TOOLS AND APPLICATIONS
(2023)
Article
Computer Science, Information Systems
Tim Hansmeier, Marco Platzner, Md Jubaer Hossain Pantho, David Andrews
JOURNAL OF SIGNAL PROCESSING SYSTEMS FOR SIGNAL IMAGE AND VIDEO TECHNOLOGY
(2019)
Article
Computer Science, Information Systems
Pankaj Bhowmik, Md Jubaer Hossain Pantho, Christophe Bobda
JOURNAL OF SIGNAL PROCESSING SYSTEMS FOR SIGNAL IMAGE AND VIDEO TECHNOLOGY
(2020)
Article
Computer Science, Hardware & Architecture
Festus Hategekimana, Taylor J. L. Whitaker, Md Jubaer Hossain Pantho, Christophe Bobda
JOURNAL OF SYSTEMS ARCHITECTURE
(2020)
Article
Chemistry, Analytical
Md Jubaer Hossain Pantho, Pankaj Bhowmik, Christophe Bobda
Summary: The advancements in optical sensing imaging technology and machine learning algorithms have enhanced the ability to extract information from scenic events, but the high computational demand of convolution neural networks limits their use in remote sensing edge devices. By designing a CNN inference architecture near the sensor and utilizing attention-based pixel processing, it is possible to optimize computations and reduce dynamic power consumption.
Article
Computer Science, Artificial Intelligence
Md Jubaer Hossain Pantho, Joel Mandebi Mbongue, Pankaj Bhowmik, Christophe Bobda
Summary: This paper introduces an event camera simulator that implements a distributed computation model to identify relevant regions in an image frame and samples frame-regions only when there is a new event. It closely emulates an image processing pipeline similar to that of physical cameras and effectively emulates event vision with low overheads according to experimental results.
JOURNAL OF REAL-TIME IMAGE PROCESSING
(2022)
Proceedings Paper
Computer Science, Hardware & Architecture
Md Jubaer Hossain Pantho, Christophe Bobda
28TH IEEE INTERNATIONAL SYMPOSIUM ON FIELD-PROGRAMMABLE CUSTOM COMPUTING MACHINES (FCCM)
(2020)
Article
Computer Science, Hardware & Architecture
Md Jubaer Hossain Pantho, Christophe Bobda
IEEE TRANSACTIONS ON COMPUTER-AIDED DESIGN OF INTEGRATED CIRCUITS AND SYSTEMS
(2020)
Proceedings Paper
Computer Science, Hardware & Architecture
Pankaj Bhowmik, Md Jubaer Hossain Pantho, Sujan Saha, Christophe Bobda
2019 IEEE COMPUTER SOCIETY ANNUAL SYMPOSIUM ON VLSI (ISVLSI 2019)
(2019)
Proceedings Paper
Computer Science, Hardware & Architecture
Md Jubaer Hossain Pantho, Pankaj Bhowmik, Christophe Bobda
2019 IEEE COMPUTER SOCIETY ANNUAL SYMPOSIUM ON VLSI (ISVLSI 2019)
(2019)
Proceedings Paper
Computer Science, Software Engineering
Pankaj Bhowmik, Md Jubaer Hossain Pantho, Christophe Bobda
PROCEEDINGS OF THE 2019 56TH ACM/EDAC/IEEE DESIGN AUTOMATION CONFERENCE (DAC)
(2019)
Proceedings Paper
Computer Science, Theory & Methods
Festus Hategekimana, Joel Mandebi Mbongue, Md Jubaer Hossain Pantho, Christophe Bobda
2018 INTERNATIONAL CONFERENCE ON FIELD-PROGRAMMABLE TECHNOLOGY (FPT 2018)
(2018)
Proceedings Paper
Computer Science, Theory & Methods
Md Jubaer Hossain Pantho, Joel Mandebi Mbongue, Christophe Bobda, David Andrews
2018 INTERNATIONAL CONFERENCE ON FIELD-PROGRAMMABLE TECHNOLOGY (FPT 2018)
(2018)
Proceedings Paper
Computer Science, Artificial Intelligence
Pankaj Bhowmik, Jubaer Hossain Pantho, Marjan Asadinia, Christophe Bobda
PROCEEDINGS 2018 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION WORKSHOPS (CVPRW)
(2018)
Proceedings Paper
Computer Science, Hardware & Architecture
Festus Hategekimana, Joel Mandebi Mbongue, Md Jubaer Hossain Pantho, Christophe Bobda
PROCEEDINGS 26TH IEEE ANNUAL INTERNATIONAL SYMPOSIUM ON FIELD-PROGRAMMABLE CUSTOM COMPUTING MACHINES (FCCM 2018)
(2018)