A5100328344- Industrial Vision Systems and Defect Detection
- Surface Roughness and Optical Measurements
- Non-Destructive Testing Techniques
- Infrastructure Maintenance and Monitoring
- Optical measurement and interference techniques
- Animal Vocal Communication and Behavior
- Traffic control and management
- Optical Network Technologies
- Robotic Path Planning Algorithms
- Advanced Neural Network Applications
- Acoustic Wave Resonator Technologies
- Blind Source Separation Techniques
- Advanced Manufacturing and Logistics Optimization
- Integrated Circuits and Semiconductor Failure Analysis
- Speech and Audio Processing
- Music and Audio Processing
Hohai University
2024
Zhengzhou University of Aeronautics
2024
Xi'an Polytechnic University
2024
Southwest Forestry University
2023
Guilin University of Technology
2022
Chongqing Jiaotong University
2021
University of Chinese Academy of Sciences
2020-2021
Shenyang Institute of Computing Technology (China)
2020-2021
Chinese Academy of Sciences
2020-2021
Automatic detection of steel surface defects is very important for product quality control in the industry. However, traditional method cannot be well applied production line, because its low accuracy and slow running speed. The current, popular algorithm (based on deep learning) also has problem accuracy, there still a lot room improvement. This paper proposes combining improved ResNet50 enhanced faster region convolutional neural networks (faster R-CNN) to reduce average time improve...
Noisy images will inevitably exist in the image acquisition process of aluminum profile surface (APS) due to limitations imaging system. And, overlap different types noise increase difficulty image-based detection for APS. Addressing this problem, we propose an APS defect method based on a deep self-attention mechanism under hybrid conditions. Firstly, use residual learning strategy obtain feature maps from noisy through identity mapping. This overcomes shallow convolutional neural network...
When deep learning is applied to intelligent textile defect detection, the insufficient training data may result in low accuracy and poor adaptability of varying types trained model. To address above problem, an enhanced generative adversarial network for augmentation improved fabric detection was proposed. Firstly, dataset preprocessed generate localization maps, which are combined with non-defective images input into training, helps better extract features. In addition, by utilizing a...
The extraction of species-specific calls from passive acoustic recordings is a common preliminary step in ecological analysis. But for many species, especially those occupying noisy, acoustically variable habitats, the call process remains largely manual, time-consuming, and increasingly unsustainable process. Deep neural networks have been shown to provide excellent performance range classification applications. We take as an example recognition four songs one rarest mammals world, western...
Abstract In a realistic scenario where large number of workpieces need to be measured, any measurement method that can detect roughness only for single workpiece is very limited in terms efficiency. To address this problem, multi-object surface detection model based on Faster R-CNN proposed paper. The features milled images with convolutional neural network. And the obtained will feed into Region Proposal Network inferring those regions may present. and go through ROI pooling layer predictor...
Existing roughness machine vision inspection methods based on designed indices require the artificial construction of features associated with image information. Such are less universal but highly demanding for imaging environment and cannot be directly applied to general industrial environments. The traditional deep learning, other hand, rely heavily a large number training samples, model needs retrained when recognition new categories is required by task. Therefore, method milling surface...
Abstract Aircraft taxiing constitutes a pivotal component of airport operations, and curtailing duration is primary strategy for enhancing operational efficiency. To actualize efficient operations alter the current state relying solely on manual apron control, this study, given unique environment aircraft taxiing, employs improved SARSA algorithm to plan path aircraft. Through simulation, it verified that surpasses traditional in terms length planning iteration times, thereby providing...
Discretizing speech into tokens and generating them by a decoder-only model have been promising direction for text-to-speech (TTS) spoken language modeling (SLM). To shorten the sequence length of tokens, acoustic byte-pair encoding (BPE) has emerged in SLM that treats from self-supervised semantic representations as characters to further compress token sequence. But gain TTS not fully investigated, proper choice BPE remains unclear. In this work, we conduct comprehensive study on various...
In the process of strip production, it is easy to produce various defects, which affect quality steel. Therefore, there are many researchers studying how use machine vision solve this problem. However, accuracy traditional learning based methods not high. paper, a method on U-net and transfer proposed. First, extend dataset using data augmentation. Then EfficientNet as backbone extract features. Use fuse features at multiple scales. Finally, four-class segmented map with size 512*512*4...