RDNet: Rotate-Groundtruth Augmentation and Decoupled Attention HEAD for 3D Object Detection

Vegetation indices, such as the normalized difference vegetation index (NDVI) or enhanced vegetation index (EVI) derived from remote sensing images, are widely used for crop classification. However, vegetation index profiles for different crops with a similar phenology lead to difficulties in discerning these crops both spectrally and temporally. This paper proposes a feature filtering and enhancement (FFE) method to map soybean and maize, two major crops widely cultivated during the summer season in Northeastern China. Different vegetation indices are first calculated and the probability density functions (PDFs) of these indices for the target classes are established based on the hypothesis of normal distribution; the vegetation index images are then filtered using the PDFs to obtain enhanced index images where the pixel values of the target classes are ”enhanced”. Subsequently, the minimum Gini index of each enhanced index image is computed, generating at the same time the weight for every index. A composite enhanced feature image is produced by summing all indices with their weights. Finally, a classification is made from the composite enhanced feature image by thresholding, which is derived automatically based on the samples. The efficiency of the proposed FFE method is compared with the maximum likelihood classification (MLC), support vector machine (SVM), and random forest (RF) in a mapping operation to determine the soybean and maize distribution in a county in Northeastern China. The classification accuracies resulting from this comparison show that the FFE method outperforms MLC, and its accuracies are similar to those of SVM and RF, with an overall accuracy of 0.902 and a kappa coefficient of 0.846. This indicates that the FFE method is an appropriate method for crop classification to distinguish crops with a similar phenology. Our research also shows that when the sample size reaches a certain level (e.g., 2000), the mean and standard deviation of the sample are very close to the actual values, which leads to high classification accuracy. In a case where the condition of normal distribution is not fulfilled, the PDF of the vegetation index can be created by a lookup table. Furthermore, as the method is rather simple and explicit, and convenient in terms of computing, it can be used as the backbone for automatic crop mapping operations.

With great significance in military and civilian applications, the topic of detecting small and densely arranged objects in wide-scale remote sensing imagery is still challenging nowadays. To solve this problem, we propose a novel effectively optimized one-stage network (NEOON). As a fully convolutional network, NEOON consists of four parts: Feature extraction, feature fusion, feature enhancement, and multi-scale detection. To extract effective features, the first part has implemented bottom-up and top-down coherent processing by taking successive down-sampling and up-sampling operations in conjunction with residual modules. The second part consolidates high-level and low-level features by adopting concatenation operations with subsequent convolutional operations to explicitly yield strong feature representation and semantic information. The third part is implemented by constructing a receptive field enhancement (RFE) module and incorporating it into the fore part of the network where the information of small objects exists. The final part is achieved by four detectors with different sensitivities accessing the fused features, all four parallel, to enable the network to make full use of information of objects in different scales. Besides, the Focal Loss is set to enable the cross entropy for classification to solve the tough problem of class imbalance in one-stage methods. In addition, we introduce the Soft-NMS to preserve accurate bounding boxes in the post-processing stage especially for densely arranged objects. Note that the split and merge strategy and multi-scale training strategy are employed in training. Thorough experiments are performed on ACS datasets constructed by us and NWPU VHR-10 datasets to evaluate the performance of NEOON. Specifically, 4.77% and 5.50% improvements in mAP and recall, respectively, on the ACS dataset as compared to YOLOv3 powerfully prove that NEOON can effectually improve the detection accuracy of small objects in remote sensing imagery. In addition, extensive experiments and comprehensive evaluations on the NWPU VHR-10 dataset with 10 classes have illustrated the superiority of NEOON in the extraction of spatial information of high-resolution remote sensing images.

Single Shot MultiBox Detector (SSD) is an effective method for multi-targets detection. However, due to the small scale and less information, small targets detection is a difficult assignment for SSD. In order to address this issue, we design an framework based on SSD, namely feature fusion and enhancement for SSD. First, considering that the feature information contained in different scale feature maps is different, especially the shallow feature maps which contains more detailed information and lacks semantic information, we propose a method of fusing different scale feature maps to introduce abundant object information. Secondly, although the feature maps has multiple channels, the importance of different channels' feature to the detected object is different. We introduce the feature enhancement module for effective feature enhancement. Training on the PASCAL VOC2007 and VOC2012, the proposed method has an mAP of 79.07%, which is 1.4% higher than SSD, and an FPS of 39 at the input size of 300 x300.

With the development of theories and technologies in medical imaging, most of the tumors can be detected in the early stage. However, the nature of ovarian cysts lacks accurate judgement, leading to that many patients with benign nodules still need Fine Needle Aspiration (FNA) biopsies or surgeries, increasing the physical pain and mental pressure of patients as well as unnecessary medical health care costs. Therefore, we present an image diagnosis system for classifying the ovarian cysts in color ultrasound images, which novelly applies the image features fused by both high-level features from deep learning network and low-level features from texture descriptor. Firstly, the ultrasound images are enhanced to improve the quality of training data set and the rotation invariant uniform local binary pattern (ULBP) features are extracted from each of the images as the low-level texture features. Then the high-level deep features extracted by the fine-tuned GoogLeNet neural network and the low-level ULBP features are normalized and cascaded as one fusion feature that can represent both the semantic context and the texture patterns distributed in the image. Finally, the fusion features are input to the Cost-sensitive Random Forest classifier to classify the images into "malignant" and "benign". The high-level features extracted by the deep neural network from the medical ultrasound image can reflect the visual features of the lesion region, while the low-level texture features can describe the edges, direction and distribution of intensities. Experimental results indicate that the combination of the two types of features can describe the differences between the lesion regions and other regions, and the differences between lesions regions of malignant and benign ovarian cysts.

A uniform transformer-based structure for feature fusion and enhancement for RGB-D saliency detection

A new method for linear feature and junction enhancement in 2D images based on morphological operation, oriented anisotropic Gaussian function and Hessian information

Fall detection based on radar technology has recently attracted increasing attention. Human motions are often multidirectional and have complex trajectories. Numerous time–frequency representation methods have been used to extract micro-Doppler fall features. Peak-to-peak Doppler frequency measurements are limited due to the relatively weak energy in the high-Doppler-frequency area. Fall features are not obvious when using a single millimeter-wave (mmWave) radar if human motions are nearly perpendicular to the radar. A fall feature enhancement and fusion method using the Stockwell transform with dual mmWave radars is proposed in this article. Two mmWave radars synchronously collect human motions in different directions, and the micro-Doppler features of these human motions are obtained by using the Stockwell transform. To enhance the features obtained for recognition, the ridgelines extracted from the two radars are fused in the time–frequency domain. In addition, a model trained by a convolutional neural network (CNN) is used for fall recognition; 960 fall motions and 720 nonfall motions are collected from 60 volunteers. The experimental results show that the proposed method can produce a recognition accuracy of up to 94.14%, which is beneficial for the realization of an indoor fall detection system.

Effectively capturing the multi-scale, multi-channel, and boundary features of target images is beneficial for improving the accuracy of salient object detection. A multichannel feature enhancement module (MFEM) is constructed by concatenating two feature maps along the channel, performing convolution activation, dot product addition, and convolution operations. An edge feature enhancement module (EFEM) is constructed by performing an inverse attention mechanism and Gaussian-Laplace transform on the predicted map, channel-wise concatenation, and convolution operations with the feature submap. Multi-scale features are extracted using a SwinTransformer encoder and enhanced using the MFEM. The enhanced features are then fused using a cross-attention mechanism for multi-level features. The fused feature map, the high-frequency feature map obtained by Gaussian-Laplace transforming the predicted map, and the external edge features are input into the EFEM to obtain the salient object detection results. This results in a novel salient object detection model with multi-scale, multi-channel, and boundary feature enhancement: MfENet. Experimental results show that the Mean Absolute Error (MAE), S-measure and F-measure accuracies of our method on the HKU-IS dataset are 0.018,0.924 and 0.959 respectively. These results demonstrate the effectiveness of our method. Furthermore, this work highlights the importance of multi-scale and boundary awareness in enhancing salient object detection, and its contribution to the technical field relevant to the theme of this conference.

The high sensitivity of the miniature mass spectrometer plays an irreplaceable role in rapid on-site detection. However, its analysis accuracy and stability should be improved due to the influence of sample pretreatment and use environment. The present study investigates the processing effects of ensemble empirical mode decomposition (EEMD) feature enhancement methods on the determination coefficient (R2 ) and relative standard deviation (RSD) of caffeine mass spectrometry signals. This paper employs the EEMD method combined with polynomial curve fitting to enhance the characteristics of seven Caffeine mass spectrum signals with different concentrations and fifteen groups of Caffeine mass spectrum signals with the same concentration, and the wavelet analysis method was used for comparative verification. The determination coefficient and relative standard deviation of the two methods were compared. We found the EEMD method's capability in adaptively decomposing Caffeine mass spectrum signals is better than wavelet analysis method. The determination coefficient of the EEMD enhanced feature is better than 0.999, and the relative standard deviation is better than 2%, and both are better than wavelet analysis methods. The feature enhancement processing using the EEMD method has significantly improved the determination coefficient and relative standard deviation of the sample curve, improving the accuracy and stability of the data and providing a new way for miniature mass spectrometer signal processing.

Image retrieval is one of the most interesting and fastest growing research areas in all fields. It is an effective and efficient tool for managing large image databases. In most Content-Based Image Retrieval (CBIR) systems, an image is represented by a set of low-level visual features; hence a direct correlation with high-level semantic information will be absent. Therefore, a gap exists between high-level information and low-level features, which is the main reason that hinders the improvement of the image retrieval accuracy. In this work, main focus is on the semantic based image retrieval system using Gray Level Co-occurrence Matrix (GLCM) for texture feature extraction. Based on the texture features, semantic interpretations are given to the extracted textures. The images are retrieved according to user satisfaction and thereby reduce the semantic gap between low level features and high level features.

This study was to analyze the clinical application value of magnetic resonance imaging (MRI) image features based on intelligent algorithms in the diagnosis and treatment of breast cancer and to provide an effective reference assessment for breast cancer diagnosis. The MRI diagnosis model (ACO-MRI) based on the ant colony algorithm (ACO) was proposed, which was compared with the diagnosis methods based on support vector machine (SVM) and proximity (KNN) algorithm, and the proposed algorithm was applied to MRI images to diagnose breast cancer. The results showed that the accuracy, sensitivity, and specificity of the ACO-MRI model were greater than those of the KNN and SVM algorithm. Moreover, the specificity was statistically considerable compared with the two algorithms of KNN and SVM ( P < 0.05 ). By comparing 1/5 number of ants and the average gray path of the ACO-MRI model under 1/8 number of ants, it was found that the average gray path value of 1/8 number of ants was greatly higher than the average gray path value of 1/5 number of ants ( P < 0.05 ). The differences in the overall distribution of breast MRI imaging features among Luminal A, Luminal B, HER-2 overexpression, and TN were compared. There were considerable differences in the overall distribution of the three breast MRI imaging features of the boundaries, morphology, and enhancement methods among the four groups ( P < 0.05 ). In short, MRI image based on the intelligent algorithm ACO-MRI diagnosis model can effectively improve the diagnosis effect of breast cancer. Its image feature boundaries, morphology, and enhancement methods had good imaging features in the diagnosis of breast cancer.

The parallel arc fault is one of the most important causes of electrical fire. However, there is no effective detection method for parallel arc fault. To solve this problem, a fault feature enhancement method based on energy transformation and reconstruction of modal components and a fault feature extraction method based on non-negative matrix factorization were proposed. First, the parallel arc fault experiments under different load current and current-limiting resistor conditions were carried out in the three-phase motor and frequency converter load circuit. Second, the current signal was decomposed into <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">k</i> modal components by using adaptive empirical wavelet transform, and the energy transformation and reconstruction of each modal component were performed in turn to enhance the fault features. Third, the parallel arc fault features were extracted from the enhancement signals by using two-stage non-negative matrix factorization dimension reduction. Finally, the performance of the fault feature enhancement and extraction method was tested by using the identification model established with a least squares support vector machine. The results indicated that the proposed method can effectively enhance and extract the parallel arc fault features from the current signals, and the detection accuracy of the parallel arc fault can reach more than 95%.

Cantonese embroidery images are difficult to recognize accurately because of significant differences in their morphologies and complex backgrounds. In this paper, a dedicated Cantonese embroidery image dataset is constructed, and an improved YOLOv8 detection method is proposed for the accurate detection of objects in Cantonese embroidery images. By integrating a feature enhancement module and the large separable kernel attention (LSKA) mechanism, the method strengthens its capacity for identifying key texture regions, enhances feature extraction in complex backgrounds, and optimizes performance through lightweight design and the WIoU loss function. The experimental results show that the method achieves a detection accuracy of 98.5%, with mAP and F1 scores improved by 0.7% and 1.37%, respectively, compared with those of the original model, and the model size is reduced by 5.12%. This provides an effective solution for the automated detection of intangible cultural heritage embroidery and the recognition of other heritage artifacts.

Recent self-supervised stereo matching methods have made significant progress. They typically rely on the photometric consistency assumption, which presumes corresponding points across views share the same appearance. However, this assumption could be compromised by real-world disturbances, resulting in invalid supervisory signals and a significant accuracy gap compared to supervised methods. To address this issue, we propose SMFormer, a framework integrating more reliable self-supervision guided by the Vision Foundation Model (VFM) and data augmentation. We first incorporate the VFM with the Feature Pyramid Network (FPN), providing a discriminative and robust feature representation against disturbance in various scenarios. We then devise an effective data augmentation mechanism that ensures robustness to various transformations. The data augmentation mechanism explicitly enforces consistency between learned features and those influenced by illumination variations. Additionally, it regularizes the output consistency between disparity predictions of strong augmented samples and those generated from standard samples. Experiments on multiple mainstream benchmarks demonstrate that our SMFormer achieves state-of-the-art (SOTA) performance among self-supervised methods and even competes on par with supervised ones. Remarkably, in the challenging Booster benchmark, SMFormer even outperforms some SOTA supervised methods, such as CFNet.

How to properly involve text characteristics like multi-scale, arbitrary direction, length aspect ratio, into detection network design has become a hot topic in computer vision. Feature Pyramid Network (FPN) is a typical method to achieve robust text detection, where its low-level and high-level feature map retains spatial structure and global semantic information, respectively. However, its strict hierarchical structure fails to fuse low-level and high-level information to improve the distinguish ability of feature map. To address this problem, we propose a novel feature fusion pyramid network for end-to-end scene text detection by fusing multi-modal information. By dividing pyramid structure into high-level and low-level layers, channel and spatial attention modules are adopted to enhance high-level and low-level feature representation by encoding channel and spatial-wise context information, respectively. In order to reduce information loss by layer transmission, a special residual network is designed to achieve short-cut between high-level and low-level features, so as to realize multi-modal feature fusion. Experiments show the precision and recall of the proposed method on ICDAR2015, ICDAR2017-MLT, and MSRA-TD500 datasets reach 88.7%/82.1%, 77.0%/60.3%, and 85.3%/74.8%, respectively.