Effectively monitoring ships and discovering abnormal ship trajectory in time is necessary for marine traffic supervision. The basic work of discovering the ship's abnormal trajectory is to predict the ship's ...
详细信息
Real-world applications of stereo matching, such as autonomous driving, place stringent demands on both safety and accuracy. However, learning-based stereo matching methods inherently suffer from the loss of geometric...
详细信息
Many daily applications are generating massive amount of data in the form of stream at an ever higher speed, such as medical data, clicking stream, internet record and banking transaction, etc. In contrast to the trad...
详细信息
Convolutional Neural Networks (CNNs)-guided deep models have obtained impressive performance for image representation, however the representation ability may still be restricted and usually needs more epochs to make t...
详细信息
ISBN:
(纸本)9781665423991
Convolutional Neural Networks (CNNs)-guided deep models have obtained impressive performance for image representation, however the representation ability may still be restricted and usually needs more epochs to make the model converge in training, due to the useful information loss during the convolution and pooling operations. We therefore propose a general feature recovery layer, termed Low-rank Deep Feature Recovery (LDFR), to enhance the representation ability of the convolutional features by seamlessly integrating low-rank recovery into CNNs, which can be easily extended to all existing CNNs-based models. To be specific, to recover the lost information during the convolution operation, LDFR aims at learning the low-rank projections to embed the feature maps onto a low-rank subspace based on some selected informative convolutional feature maps. Such low-rank recovery operation can ensure all convolutional feature maps to be reconstructed easily to recover the underlying subspace with more useful and detailed information discovered, e.g., the strokes of characters or the texture information of clothes can be enhanced after LDFR. In addition, to make the learnt low-rank subspaces more powerful for feature recovery, we design a fusion strategy to obtain a generalized subspace, which averages over all learnt sub-spaces in each LDFR layer, so that the convolutional feature maps in test phase can be recovered effectively via low-rank embedding. Extensive results on several image datasets show that existing CNNs-based models equipped with our LDFR layer can obtain better performance.
High junction temperature (Tj) of LED will affect the performance and life of LED, so this paper proposes a high-power LED junction temperature test method. Mainly aiming at the problem that the accuracy of the forwar...
详细信息
This paper presents a hybrid method using a Gaitset network for gait recognition. We firstly use Alphpose model to obtain key points of human posture to improve the extraction of human contour under body occlusion. Th...
详细信息
Multi-dimensional classification (MDC) has attracted much attention from the community. Though most studies consider fully annotated data, in real practice obtaining fully labeled data in MDC tasks is usually intracta...
High-resolution magnetic resonance imaging (MRI) provides a clear anatomical structure for diagnosis, however, its high cost makes it unsuitable in practice. On the contrary, low-resolution MRI cannot provide fine str...
High-resolution magnetic resonance imaging (MRI) provides a clear anatomical structure for diagnosis, however, its high cost makes it unsuitable in practice. On the contrary, low-resolution MRI cannot provide fine structural information but is resource and time efficient. Super-resolution methods are high-resolution images obtained from low-resolution MRI. However, existing MRI super-resolution methods suffer from the following defects: (1) CNN-based super-resolution methods lack the ability to build relationships for remote features. (2) While models using transformer based on self-attentive mechanism have achieved a breakthrough in super-resolution, the single self-attentive model fails to fully utilize the useful features of the input image. Thus, in this paper, an MRI super-resolution network (MATNet) based on a multi-attention mechanism is proposed that can utilize multiple useful image features from different dimensions for image super-resolution, improved super-resolution performance. To enable the network to obtain noise-free image features before sampling the image, we also added a denoising module to it. Finally, we prove the effectiveness of the model using the IXI public dataset.
This paper studies the multimedia problem of temporal sentence grounding (TSG), which aims to accurately determine the specific video segment in an untrimmed video according to a given sentence query. Traditional TSG ...
详细信息
Most of the unsupervised hashing methods usually map images into semantic similarity-preserving hash codes by constructing local semantic similarity structure as guiding information, i.e., treating each point similar ...
详细信息
暂无评论