A5075392349- Advanced Neural Network Applications
- Advanced Image and Video Retrieval Techniques
- Video Surveillance and Tracking Methods
- Domain Adaptation and Few-Shot Learning
- Brain Tumor Detection and Classification
- IoT-based Smart Home Systems
- COVID-19 diagnosis using AI
- CCD and CMOS Imaging Sensors
Tamkang University
2022-2023
With the continuous development of neural networks in computer vision tasks, more and network architectures have achieved outstanding success. As one most advanced architectures, DenseNet shortcuts all feature maps to solve problem model depth. Although this architecture has excellent accuracies with low parameters, it takes excessive inference time. To problem, HarDNet reduces connections between maps, making remaining resemble harmonic waves. However, compression method may result...
This paper proposes a new pure attention model, Aggregated Pyramid Vision Transformer (APVT), for computer vision applications. Based on the (ViT) architecture, APVT adopts classic pyramid architecture of CNN and employs group encoder technique to replace traditional feature enhancement. uses split-transform-merge strategy refine operation. The model performs image classification CIFAR-10 dataset object detection COCO 2017 verification. Experimental results show that has excellent...
With the achievements of Transformer in field natural language processing, encoder-decoder and attention mechanism have been applied to computer vision. Recently, multiple tasks vision (image classification, object detection, semantic segmentation, etc.), state-of-the-art convolutional neural networks introduced some concepts Transformer. This proves that has a good prospect image recognition. After Vision was proposed, more works began use self-attention completely replace layer. work is...
With the good performance of deep learning algorithms in field computer vision (CV), convolutional neural network (CNN) architecture has become a main backbone task. widespread use mobile devices, models based on platforms with low computing power are gradually being paid attention. However, due to limitation power, usually not available devices. This paper proposes lightweight network, TripleNet, which can operate easily Raspberry Pi. Adopted from concept block connections ThreshNet, newly...