Evolutionary algorithms (EAs) have been applied successfully in many fields. However, EAs cannot find an optimal solution on occasions because the balance between exploration and exploitation is lost runs. So far, tricking important research topic field of evolutionary computation. Elitism strategy a typical scheme selection for above purpose can be widely used different EAs. In this paper, we propose elitism distance based strategy. According to our strategy, elites are still kept reducing...

Glass products are important evidence of early East–West cultural exchanges. Ancient glass in China mostly consisted lead glass, and potassium is widely believed to be imported abroad. In order figure out the origin artefacts, it crucial define type accurately. contemporary research on chemical composition ancient products, separated from primarily by weight ratio oxides or proportion lead-containing compounds. This approach can excessively subjective prone mistakes while calculating mass...

The aim of this research is to show the implementation object detection on drone videos using TensorFlow API. function recognition effect and performance popular target algorithm feature extractor for recognizing people, trees, cars, buildings from real-world video frames taken by drones. study found that different algorithms “normal” image (an ordinary camera) has effects number instances, accuracy, consumption application data acquired different. Object a key part realization any robot’s...

At present, there is a growing emphasis on safeguarding image data, yet conventional encryption methods are full of numerous limitations. In order to tackle the limitations color methodologies, such as inefficiency and insufficient security, this paper designs an expedited method for images that uses DNA coding in conjunction with multiple chaotic systems. The algorithm proposed based three-dimensional permutation, global scrambling, one-dimensional diffusion coding. First all, permutation...

In modern manufacturing, cameras are widely used to record the full workflow of assembly line workers, enabling video-based operational analysis and management. However, these recordings often excessively long, leading high storage demands inefficient processing. Existing keyframe extraction methods typically apply uniform strategies across all frames, which ineffective in detecting subtle movements. To address this, we propose a method tailored for videos, combining optical flow estimation...

Statistical analysis and research on insect grooming behavior can find more effective methods for pest control. Traditional manual statistical are time-consuming, labor-intensive, error-prone. Based computer vision technology, this paper uses spatio-temporal context to extract video features, self-built Convolution Neural Network (CNN) train the detection model, proposes a simple Bactrocera minax method, which automatically detects behaviors of flies results by program. Applying method...

Moving vehicles detection is an important part of Intelligence Transport System (ITS). Traditional method moving from video image subtraction which apt to be affected by brightness changing, in this paper a kind algorithm based on optical flow purposed: estimate through two consecutive frames' pyramids and compute every image's threshold with images are segment into binaryzation images, after that, morphological transformation operator rectangular splitting vehicles' will extracted...

Abstract Background Compared with traditional surgical methods, which only rely on doctors' experience, three‐dimensional robot‐assisted navigation technology could locate the lesion more accurately. Methods The transformation principle of system was analyzed. path planned reconstructed model. experimental data positioning position and posture were used as two important parameters to evaluate accuracy spine surgery system. Results final results show that 2.54 ± 0.15 mm (range 2.25‐2.76 mm)....

Deep convolutional neural networks is playing very important role to solve computer vision task in these decades. In this research has shown implementation state of art face recognition methods and compared them. Advantages disadvantages Convolutional Siamese explored for the task. The novel system accuracy checked on widely used Labelled Faces Wild (LFW) dataset. study introduced concepts triplet loss function. Experiment results can be useful developers who going create AI applications...

Statistical analysis of Bactrocera grooming behavior is important for pest control and human health. Based on DeepLabCut, this study proposes a noninvasive effective method to track the key points minax detect analyze its behavior. The results are analyzed calculated automatically by computer program. Traditional movement tracking methods invasive; instance, use artificial pheromone may affect minax, thus directly affecting accuracy reliability experimental results. research studies mainly...

The reading of the ship draft is an important step in process weighing and pricing. traditional detection method time-consuming labor-consuming, it easy to lead misdetection. In order solve above problems, this paper introduces computer image processing technology based on deep learning, specific divided into three steps: first, video sampling carried out by UAV obtain a large number pictures face, images are preprocessed; then, learning target algorithm improved YOLOv3 used predict position...

Pest management has long been a critical aspect of crop protection. Insect behavior is great research value as an important indicator for assessing insect characteristics. Currently, increasingly based on the quantification behavior. Traditional manual observation and analysis methods can no longer meet requirements data volume time. In this paper, we propose method region localization combined with improved 3D convolutional neural network six grooming behaviors Bactrocera minax: head...

Identifying and statistically analyzing soybean pod types are crucial for seed evaluation yield estimation. Traditional visual assessment by breeding personnel is time-consuming, labor-intensive, prone to subjective bias, especially with large datasets. Automatic methods usually struggle the highly confusing two three seeds, affecting model’s identification accuracy. To address these issues, we propose improve standard YOLOv5s object detection model enhance differentiation between boost...

With technology of digital image recognition, the data frames include vehicle’s can be detected automatically from an Audio Video Interleaved (AVI) File, and vehicle type will recognized automatically. The system consists of four modules: Reading AVI file decomposing it into frames; Motion detection; Vehicle’s processing; Module classification. In particularly, algorithm recognition through counting number black pixels included in body contour is one innovation algorithm. Experiment on...

The single-person pose estimation algorithm based on OpenPose and the multi-person detection are used to process action image of person collected by camera realize positioning key points human face, hand, various joints body points; Bresenham is draw lines body. accuracy traditional posture recognition will be reduced when it recognizes occluded parts. This paper improves problem algorithm. experimental results show that method adopted in this can adapt complex background, has higher operation rate.