Related references
Note: Only part of the references are listed.
Article
Chemistry, Analytical
Rujing Wang et al.
Summary: The introduction of a domain-specific benchmark dataset called AgriPest aims to provide standard large-scale dataset in wild pest recognition and detection, enabling researchers and communities to access practical wild pest images and annotations for evaluation. AgriPest captures 49.7K images of four crops containing 14 species of pests over the past seven years, manually annotated by agricultural experts with up to 264.7K bounding boxes. The detailed analysis and evaluation of AgriPest offers great opportunities for researchers in computer vision and pest monitoring applications.
Article
Chemistry, Analytical
Yawen Luo et al.
Summary: Additive manufacturing (AM) offers advantages like fast prototyping, easy modification, and complex internal structures but requires real-time defect inspection. Combining training with NEU-DET, convolution neural networks, FPGA acceleration, and binarized neural networks, real-time defect inspection with high accuracy is achieved. The modular design allows for easy scaling to larger FPGA implementations.
Article
Chemistry, Analytical
Liyun Xiao et al.
Summary: The study proposes an improved YOLOv3-based multi-directional text detection algorithm, which introduces sliding vertices box definition and a new rotating box loss function to improve detection accuracy, with further reduction of computation using a step-by-step NMS method. Experimental results show good detection performance on natural scene text detection in the ICDAR 2015 dataset.
Article
Chemistry, Analytical
Xiaofang Zhao et al.
Summary: An improved DETR object detection framework is proposed, which uses ResNeSt backbone network and multi-scale location encoding to increase the accuracy of detection and recognition of characters on shipping containers, achieving an overall accuracy of 98.6%.
Article
Chemistry, Analytical
Ui-Nyoung Yoon et al.
Summary: This paper introduces an unsupervised video summarization method, Interp-SUM, utilizing piecewise linear interpolation for generating a natural sequence of keyframes. Through reinforcement learning-based training framework, it achieves competitive performance compared to state-of-the-art methods in experiments.
Article
Chemistry, Analytical
Fenglong Ding et al.
Summary: An optimized convolutional neural network was proposed to identify the tree species of sawn timber. Through multiple comparative experiments, it was found that the prediction model using a linear function as the kernel function of support vector machine learning the feature vectors from the improved convolution layer performed best, with an overall accuracy of all samples above 99%.
Article
Chemistry, Analytical
Jie Kou et al.
Summary: The study proposed a new spark detection network SAVSDN, which effectively differentiates between sparks and interference by integrating video codec and image object detection methods. By generating the simulated spark image dataset SAES to address the lack of actual spark data, the experiments demonstrated that SAVSDN performs exceptionally well in spark detection.
Article
Chemistry, Analytical
Taimur Hassan et al.
