Fully convolutional neural networks (FCNs) have shown outstanding performance in many dense labeling problems. One key pillar of these successes is mining relevant information from features layers. However, how to better aggregate multi-level feature maps for salient object detection underexplored. In this work, we present Amulet, a generic aggregating framework detection. Our first integrates into multiple resolutions, which simultaneously incorporate coarse semantics and fine details. Then...

Though deep learning techniques have made great progress in salient object detection recently, the predicted saliency maps still suffer from incomplete predictions due to internal complexity of objects and inaccurate boundaries caused by strides convolution pooling operations. To alleviate these issues, we propose train networks exploiting supervision not only detection, but also foreground contour edge detection. First, leverage tasks an intertwined manner generate with uniform highlight....

Multi-Object Tracking (MOT) aims to detect and associate all desired objects across frames. Most methods accomplish the task by explicitly or implicitly leveraging strong cues (i.e., spatial appearance information), which exhibit powerful instance-level discrimination. However, when object occlusion clustering occur, information will become ambiguous simultaneously due high overlap among objects. In this paper, we demonstrate long-standing challenge in MOT can be efficiently effectively...

Fully convolutional neural networks (FCNs) have shown outstanding performance in many dense labeling problems. One key pillar of these successes is mining relevant information from features layers. However, how to better aggregate multi-level feature maps for salient object detection underexplored. In this work, we present Amulet, a generic aggregating framework detection. Our first integrates into multiple resolutions, which simultaneously incorporate coarse semantics and fine details. Then...

Automatic seizure prediction promotes the development of closed-loop treatment system on intractable epilepsy. In this study, by considering specific information exchange between EEG channels from perspective whole brain activities, convolution neural network (CNN) and directed transfer function (DTF) were merged to present a novel method for patient-specific prediction. Firstly, intracranial electroencephalogram (iEEG) signals segmented flow features iEEG calculated using DTF algorithm....

Traffic sign detection and recognition plays an important role in expert systems, such as traffic assistance driving systems automatic systems. It instantly assists drivers or detecting recognizing signs effectively. In this paper, a novel approach for real-time real situation was proposed. First, the images of road scene were converted to grayscale images, then we filtered with simplified Gabor wavelets (SGW), where parameters optimized. The edges strengthened, which helpful next stage...

Traffic sign detection systems provide important road control information for unmanned driving or auxiliary driving. In this paper, the Faster region with a convolutional neural network (R-CNN) traffic in real situations has been systematically improved. First, first step proposal algorithm based on simplified Gabor wavelets (SGWs) and maximally stable extremal regions (MSERs) is proposed. way, priori obtained will be used improving R-CNN. This part of our method named as highly possible...

The precise identification of retinal disorders is utmost importance in the prevention both temporary and permanent visual impairment. Prior research has yielded encouraging results classification images pertaining to a specific condition. In clinical practice, it not uncommon for single patient present with multiple concurrently. Hence, task classifying into labels remains significant obstacle existing methodologies, but its successful accomplishment would yield valuable insights diverse...

To realize the fault diagnosis of bearing effectively, this paper presents a novel method based on Gaussian restricted Boltzmann machine (Gaussian RBM). Vibration signals are firstly resampled to same equivalent speed. Subsequently, envelope spectrums data used directly as feature vectors represent types bearing. Finally, in order deal with high-dimensional spectrum, classifier model RBM is applied. has ability provide closed-form representation distribution underlying training data, and it...

Existing video captioning methods usually ignore the important fine-grained semantic attributes, diversity, as well association and motion state between objects within frames. Thus, they cannot adapt to small sample data sets. To solve above problems, this paper proposes a novel model an adversarial reinforcement learning strategy. Firstly, object-scene relational graph is designed based on object detector scene segmenter express features. The encoded by neural network enrich expression of...

With the aim of automatic recognition weak faults in hydraulic systems, this paper proposes an identification method based on multi-scale permutation entropy feature extraction fault-sensitive intrinsic mode function (IMF) and deep belief network (DBN). In method, leakage fault signal is first decomposed by empirical decomposition (EMD), IMF components are screened adopting correlation analysis method. The each then extracted features closely related to information obtained. Finally, DBN...

Deep-learning convolutional neural networks (CNNs) have proven to be successful in various cognitive applications with a multilayer structure. The high computational energy and time requirements hinder the practical application of CNNs; hence, realization highly energy-efficient fast-learning network has aroused interest. In this work, we address computing-resource-saving problem by developing deep model, termed Gabor (Gabor CNN), which incorporates expression-efficient kernels into CNNs....

This paper proposes an Agile Aggregating Multi-Level feaTure framework (Agile Amulet) for salient object detection. The Amulet builds on previous works to predict saliency maps using multi-level convolutional features. Compared works, employs some key innovations improve training and testing speed while also increase prediction accuracy. More specifically, we first introduce a contextual attention module that can rapidly highlight most objects or regions with pyramids. Thus, it effectively...

Infrared target intrusion detection has significant applications in the fields of military defence and intelligent warning. In view characteristics targets as well inspection difficulties, an infrared algorithm based on feature fusion enhancement was proposed. This combines static mode analysis dynamic multi-frame correlation to extract features at different levels. Among them, LBP texture can be used effectively identify posterior patterns which have been contained library, while motion...

Visual object tracking is a fundamental video task in computer vision. Recently, the notably increasing power of perception algorithms allows unification single/multiobject and box/mask-based tracking. Among them, Segment Anything Model (SAM) attracts much attention. In this report, we propose HQTrack, framework for High Quality Tracking anything videos. HQTrack mainly consists multi-object segmenter (VMOS) mask refiner (MR). Given to be tracked initial frame video, VMOS propagates masks...

Transformer-based visual trackers have demonstrated significant progress owing to their superior modeling capabilities. However, existing are hampered by low speed, limiting applicability on devices with limited computational power. To alleviate this problem, we propose HiT, a new family of efficient tracking models that can run at high speed different while retaining performance. The central idea HiT is the Bridge Module, which bridges gap between modern lightweight transformers and...