Vehicle detection in aerial images is an important and challenging task. Traditionally, many target models based on sliding-window fashion were developed achieved acceptable performance, but these are time-consuming the phase. Recently, with great success of convolutional neural networks (CNNs) computer vision, state-of-the-art detectors have been designed deep CNNs. However, CNN-based inefficient when applied image data due to fact that existing struggle small-size object precise...

With the development of deep learning algorithms, more and algorithms are being applied to remote sensing image classification, detection, semantic segmentation. The landslide segmentation a based on mainly uses supervised learning, accuracy which depends large number training data high-quality annotation. At this stage, annotation often requires investment significant human effort. Therefore, high cost greatly restricts algorithm. Aiming resolve problem labeling with method, we proposed...

Multi-objective optimization is a cornerstone of modern engineering and management, tackling challenges in complex system design, resource allocation, financial portfolio optimization. Effective multi-objective algorithms must strike balance between convergence diversity, process that inherently reflects the symmetry objectives their trade-offs. However, real-world complexities introduce significant hurdles: exponential increase Pareto optimal solutions diminishes effectiveness...

This paper proposes a novel high-sensitivity micro-electromechanical system (MEMS) piezoresistive pressure sensor that can be used for rock mass stress monitoring. The entire consists of cross, dual-cavity, and all-silicon bulk-type (CCSB) structure. Firstly, the theoretical analysis is carried out, relationship between structural parameters analyzed by finite element simulation curve-fitting prediction, then optimal are also analyzed. results indicate with CCSB structure proposed in this...

Background subtraction (BS) is one of the most commonly encountered tasks in video analysis and tracking systems. It distinguishes foreground (moving objects) from sequences captured by static imaging sensors. remote scene infrared (IR) important common to lots fields. This paper provides a Remote Scene IR Dataset our designed medium-wave (MWIR) sensor. Each sequence this dataset identified with specific BS challenges pixel-wise ground truth (FG) for each frame also provided. A series...

In this paper, we propose a novel deep learning the based deraining method. The proposed method is motivated by idea that an effective algorithm should have ability to remove various remaining rain streaks, which been processed method, in repeated way. So, design network coarse-to-fine manner multi-stage processing procedure and parameters are shared each stage. As spatial contextual information important for single image deraining, densely connected dilation convolution block deal with...

Identifying wildfires is a key task to ensure timely and effective response prevent the spread of wildfires. The widely used methods for identifying based on infrared satellite remote sensing still have problems such as poor identification accuracy high deployment costs. To address costs, this paper adopted method deep learning network visible light images.To issue accuracy, proposed an intelligent wildfire Weighted Boxes Fusion (WBF) Convolutional Block Attention Module (CBAM). In order...

Many objects in the naturalenvironment are generated from background and even transformed by nature or human beings. Thus, they do not have closed well-defined boundaries remote sensing imagery. Recently, convolutional neural network (CNN) based object detection achieved great success field. However, there is no investigation literature about of with ambiguous boundaries. In this article, taking case potential loess landslide detection, we designed massive experiments to evaluate networks...

Video action recognition is widely applied in video indexing, intelligent surveil-lance, multimedia understanding, and other fields. Recently, it was greatly improved by incorporating the learning of deep information using convolutional neural network (ConvNet). In this paper, we proposed a 3D ConvNet-GRU architecture to learn for recognition. Specifically, use ConvNet spatiotemporal from short RGB clips optical flow clips, impose gated recurrent unit (GRU) on in-formation model temporal...

Wildfire, also known as forest fire, is fire that usually occur in forests and are difficult to control. If it could be detected suppressed at an early stage (mainly smoke flames), has important meaning for reducing the loss. With attention of relevant researchers, wildfire detection technology become more advanced, from traditional manual monitoring target sensor infrared detection, etc. The various methods involved still have problems such slow speed, low accuracy, easy interference high...

Temporal action detection in long, untrimmed videos is an important yet challenging task that requires not only recognizing the categories of actions videos, but also localizing start and end times each action. Recent years, artificial neural networks, such as Convolutional Neural Network (CNN) Long Short-Term Memory (LSTM) improve performance significantly various computer vision tasks, including detection. In this paper, we make most different granular classifiers propose to detect from...

Few-shot class-incremental learning (FSCIL) is crucial and practical for artificial intelligence in the real world, which learns novel classes incrementally from few samples without forgetting previously learned classes. However, FSCIL confronts two significant challenges: “catastrophic forgetting” “overfitting new.” We focus on convolutional neural network (CNN)-based propose a human cognition-inspired method, knowledge of under guidance knowledge. Specifically, we learn discriminative...

Aircraft detection is the main task of optoelectronic guiding and monitoring system in airports. In practical applications, we demand not only accuracy, but also efficiency. Existing approaches always train a set holistic templates to search over multi-scale image space, which inefficient costly. Moreover, are sensitive occluded or truncated object, although they trained by many complicated features. To address these problems, firstly propose kind local informative feature combines patch...