Direct Concatenation Research Articles

Rapid Determination of Soil Organic Carbon and Permanganate Oxidizable Carbon: Comparison of Individual and Data Fusion Performance Using Fourier Transform Mid-Infrared Photoacoustic Spectroscopy (FTIR-PAS) and Laser-Induced Breakdown Spectroscopy (LIBS)

Rapid and accurate prediction of soil organic carbon (SOC) and the carbon fractions that are sensitive to management are vital for evaluating soil fertility and SOC turnover from both long-term and short-term perspectives. This study used 396 soil samples to investigate the individual and combined use of FTIR-PAS and LIBS for both SOC and permanganate oxidizable carbon (POXC) prediction coupled with partial least squares regression (PLSR). The results showed that LIBS from 205 to 899 nm (LIBS-H) was suitable for predicting SOC and POXC, with a coefficient of determination in prediction ( R P 2 ) of 0.68 and 0.73, and a root mean square error in prediction (RMSEP) of 0.57% and 169 mg kg−1, respectively. FTIR-PAS also exhibited good potential for the prediction of SOC and POXC with a slightly poorer performance ( R P 2 of 0.66 and RMSEP of 0.59% for SOC; R P 2 of 0.72 and RMSEP of 171 mg kg−1 for POXC). The prediction ability of SOC was improved by low-level data fusion based on direct concatenation of FTIR-PAS spectra and LIBS spectra at the range of 184–205 nm (LIBS-L), resulting in an R P 2 of 0.72 and an RMSEP of 0.53%. In general, there is good potential for combining FTIR-PAS and LIBS-L to improve the SOC prediction. Both FTIR-PAS and LIBS-H can be applied for individual prediction of SOC and POXC. In addition, FTIR-PAS offers the advantages of reduced complexity in the modeling and easier sample preparation.

Mitigating imbalances in heterogeneous feature fusion for multi-class 6D pose estimation

Most 6D pose studies often treat RGB and Depth features equally in fusion, potentially limiting model generalization, especially in multi-class tasks. This limitation arises from prevalent static map generation strategies that overlook discriminative features in downsampling sparse point clouds. Additionally, the commonly adopted direct concatenation approach in heterogeneous feature fusion often leads to an averaging effect, thereby reducing the effectiveness of each feature. To tackle these challenges, we propose an effective model for dynamic graph structure feature extraction, aimed at capturing richer features from point clouds. And we introduce an adaptive fusion method for heterogeneous features, which takes into account the unequal contributions to 6D pose estimation. Validation on benchmark datasets LineMOD and YCB-Video demonstrates its effectiveness for multi-class 6D pose estimation, surpassing existing fusion methods. Of particular significance, our method attains state-of-the-art (SOTA) results on the YCB-Video dataset. The code for this study can be accessed at https://github.com/ZEROhands/6D_Pose_Estimate.

A Weighted Symmetric Graph Embedding Approach for Link Prediction in Undirected Graphs.

Link prediction is an important task in social network analysis and mining because of its various applications. A large number of link prediction methods have been proposed. Among them, the deep learning-based embedding methods exhibit excellent performance, which encodes each node and edge as an embedding vector, enabling easy integration with traditional machine learning algorithms. However, there still remain some unsolved problems for this kind of methods, especially in the steps of node embedding and edge embedding. First, they either share exactly the same weight among all neighbors or assign a completely different weight to each node to obtain the node embedding. Second, they can hardly keep the symmetry of edge embeddings obtained from node representations by direct concatenation or other binary operations such as averaging and Hadamard product. In order to solve these problems, we propose a weighted symmetric graph embedding approach for link prediction. In node embedding, the proposed approach aggregates neighbors in different orders with different aggregating weights. In edge embedding, the proposed approach bidirectionally concatenates node pairs both forwardly and backwardly to guarantee the symmetry of edge representations while preserving local structural information. The experimental results show that our proposed approach can better predict network links, outperforming the state-of-the-art methods. The appropriate aggregating weight assignment and the bidirectional concatenation enable us to learn more accurate and symmetric edge representations for link prediction.

Learning Multi-Modal Scale-Aware Attentions for Efficient and Robust Road Segmentation

Road segmentation is essential to unmanned systems, contributing to road perception and navigation in the field of autonomous driving. While multi-modal road segmentation methods have shown promising results by leveraging the complementary data of RGB and Depth to provide robust 3D geometry information, existing methods suffer from severe efficiency problems that hinder their practical application in autonomous driving. Their direct concatenation of multi-modal features with a densely-connected network leads to increased semantic gaps among modalities and scales, causing high computational and time complexity. To address these issues, we propose a Multi-modal Scale-aware Attention Network (MSAN) to fuse RGB and Depth data effectively via a novel transformer-based cross-attention module, namely Multi-modal Scare-aware Transformer (MST), which fuses RGB-D features from a global perspective across multiple scales. To better consolidate different scales of features, we further propose a Scale-aware Attention Module (SAM) that captures channel-wise attention efficiently for cross-scale fusion. These two attention-based modules explore the complementarity of modalities and scales, narrowing the gaps and avoiding complex structures for road segmentation. Extensive experiments demonstrate MSAN achieves competitive performance at a low computational cost, suitable for real-time implementation on edge-devices in autonomous driving systems.

High-Resolution Remote Sensing Image Change Detection Method Based on Improved Siamese U-Net

Focusing on problems of blurred detection boundary, small target miss detection, and more pseudo changes in high-resolution remote sensing image change detection, a change detection algorithm based on Siamese neural networks is proposed. Siam-FAUnet can implement end-to-end change detection tasks. Firstly, the improved VGG16 is utilized as an encoder to extract the image features. Secondly, the atrous spatial pyramid pooling module is used to increase the receptive field of the model to make full use of the global information of the image and obtain the multi-scale contextual information of the image. The flow alignment module is used to fuse the low-level features in the encoder to the decoder and solve the problem of semantic misalignment caused by the direct concatenation of features when the features are fused, so as to obtain the change region of the image. The experiments are trained and tested using publicly available CDD and SZTAKI datasets. The results show that the evaluation metrics of the Siam-FAUnet model are improved compared to the baseline model, in which the F1-score is improved by 4.00% on the CDD and by 7.32% and 2.62% on the sub-datasets of SZTAKI (SZADA and TISZADOB), respectively; compared to other state-of-the-art methods, the Siam-FAUnet model has improved in both evaluation metrics, indicating that the model has a good detection performance.

Visual–tactile fusion object classification method based on adaptive feature weighting

Visual–tactile fusion information plays a crucial role in robotic object classification. The fusion module in existing visual–tactile fusion models directly splices visual and tactile features at the feature layer; however, for different objects, the contributions of visual features and tactile features to classification are different. Moreover, direct concatenation may ignore features that are more beneficial for classification and will also increase computational costs and reduce model classification efficiency. To utilize object feature information more effectively and further improve the efficiency and accuracy of robotic object classification, we propose a visual–tactile fusion object classification method based on adaptive feature weighting in this article. First, a lightweight feature extraction module is used to extract the visual and tactile features of each object. Then, the two feature vectors are input into an adaptive weighted fusion module. Finally, the fused feature vector is input into the fully connected layer for classification, yielding the categories and physical attributes of the objects. In this article, extensive experiments are performed with the Penn Haptic Adjective Corpus 2 public dataset and the newly developed Visual-Haptic Adjective Corpus 52 dataset. The experimental results demonstrate that for the public dataset Penn Haptic Adjective Corpus 2, our method achieves a value of 0.9750 in terms of the area under the curve. Compared with the highest area under the curve obtained by the existing state-of-the-art methods, our method improves by 1.92%. Moreover, compared with the existing state-of-the-art methods, our method achieves the best results in training time and inference time; while for the novel Visual-Haptic Adjective Corpus 52 dataset, our method achieves values of 0.9827 and 0.9850 in terms of the area under the curve and accuracy metrics, respectively. Furthermore, the inference time reaches 1.559 s/sheet, demonstrating the effectiveness of the proposed method.

Recurrent Structure Attention Guidance for Depth Super-resolution

Image guidance is an effective strategy for depth super-resolution. Generally, most existing methods employ hand-crafted operators to decompose the high-frequency (HF) and low-frequency (LF) ingredients from low-resolution depth maps and guide the HF ingredients by directly concatenating them with image features. However, the hand-designed operators usually cause inferior HF maps (e.g., distorted or structurally missing) due to the diverse appearance of complex depth maps. Moreover, the direct concatenation often results in weak guidance because not all image features have a positive effect on the HF maps. In this paper, we develop a recurrent structure attention guided (RSAG) framework, consisting of two important parts. First, we introduce a deep contrastive network with multi-scale filters for adaptive frequency-domain separation, which adopts contrastive networks from large filters to small ones to calculate the pixel contrasts for adaptive high-quality HF predictions. Second, instead of the coarse concatenation guidance, we propose a recurrent structure attention block, which iteratively utilizes the latest depth estimation and the image features to jointly select clear patterns and boundaries, aiming at providing refined guidance for accurate depth recovery. In addition, we fuse the features of HF maps to enhance the edge structures in the decomposed LF maps. Extensive experiments show that our approach obtains superior performance compared with state-of-the-art depth super-resolution methods. Our code is available at: https://github.com/Yuanjiayii/DSR-RSAG.

Quality-Driven Dual-Branch Feature Integration Network for Video Salient Object Detection

Video salient object detection has attracted growing interest in recent years. However, some existing video saliency models often suffer from the inappropriate utilization of spatial and temporal cues and the insufficient aggregation of different level features, leading to remarkable performance degradation. Therefore, we propose a quality-driven dual-branch feature integration network majoring in the adaptive fusion of multi-modal cues and sufficient aggregation of multi-level spatiotemporal features. Firstly, we employ the quality-driven multi-modal feature fusion (QMFF) module to combine the spatial and temporal features. Particularly, the quality scores estimated from each level’s spatial and temporal cues are not only used to weigh the two modal features but also to adaptively integrate the coarse spatial and temporal saliency predictions into the guidance map, which further enhances the two modal features. Secondly, we deploy the dual-branch-based multi-level feature aggregation (DMFA) module to integrate multi-level spatiotemporal features, where the two branches including the progressive decoder branch and the direct concatenation branch sufficiently explore the cooperation of multi-level spatiotemporal features. In particular, in order to provide an adaptive fusion for the outputs of the two branches, we design the dual-branch fusion (DF) unit, where the channel weight of each output can be learned jointly from the two outputs. The experiments conducted on four video datasets clearly demonstrate the effectiveness and superiority of our model against the state-of-the-art video saliency models.

Analysis of ENF Signal Extraction From Videos Acquired by Rolling Shutters

Electric network frequency (ENF) analysis is a promising forensic technique for authenticating multimedia recordings and detecting tampering. The validity of the ENF analysis heavily relies on the capability of extracting high-quality ENF signals from multimedia recordings. This paper analyzes and compares two representative methods for extracting ENF signals from visual signals acquired by cameras using the rolling-shutter mechanism. The first method proposed in prior work, <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">direct concatenation</i> , ignores the idle period of each frame. The second method proposed in this paper, <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">periodic zeroing-out</i> , inserts zeros to missing sample points instead of ignoring the idle period. Our theoretical analyses of using multirate signal processing reveal and experiments confirm that while the first method can extract ENF signals without knowing the exact value of camera read-out time, there exists some mild distortion to extracted ENF signals. In contrast, the second method taking the read-out time as the additional input is capable of extracting distortion-free ENF signals, and its frequency component of the highest strength is always located at the nominal frequency. Additionally, we examine aliased DC and negative ENF components caused by the two methods and show that their impact on the accuracy of frequency estimation is minimum. This paper facilitates the fundamental understanding of extracting ENF signals from videos. The research findings imply that the periodic zeroing-out method offers more accurate frequency estimates, but the performance improvement is not significant.

Improving soil organic matter estimation accuracy by combining optimal spectral preprocessing and feature selection methods based on pXRF and vis-NIR data fusion

Rapid and accurate estimation of soil organic matter (SOM) content is of great significance for agricultural production and carbon stock estimation. Visible-near-infrared spectroscopy (Vis-NIR) and portable X-ray fluorescence spectroscopy (pXRF), with the fast, low-cost, and environmental friendly characteristics, can realize the rapid large-scale detection of SOM. However, due to the heterogeneity and complexity of soil, estimation of SOM using a single sensor often leads to high uncertainty in result. Multi-sensor data fusion is one of the approaches to solve this problem, but suitable spectral preprocessing methods and data fusion methods still need further research. In this study, 500 soil samples were collected from two watersheds with large differences in soil parent material, soil type, and climate in the Xinjiang Uygur Autonomous Region of China. The effects of different spectral preprocessing methods for vis-NIR and pXRF spectra on SOM estimation were compared, and the effects of five feature selection methods on single sensor estimation and three different multi-sensor fusion estimations were evaluated. The results showed that vis-NIR spectrum was better than pXRF spectrum for SOM estimation, and the 1.6th-order derivative and baseline correction were the optimal preprocessing method for vis-NIR and pXRF spectra (Beam 3), respectively. Feature selection could improve the accuracy of the single sensor and multi-sensor fusion estimations. The competitive adaptive reweighted sampling (CARS) method was the optimal feature selection method for single sensor estimation and Granger–Ramanathan averaging (GRA) fusion estimation, and the variable importance in projection (VIP) was the optimal feature selection method for direct concatenation and outer-product analysis fusion estimation. In general, the optimal SOM estimation (RMSEp = 3.59 g kg−1, LCCCp = 0.84, RPIQp = 2.67) could be achieved based on CARS feature selection method and GRA fusion. This study proves that the combination of spectral preprocessing and feature selection can effectively improve the accuracy of the single sensor and multi-sensor fusion estimations. Especially, the multi-sensor fusion estimation has higher accuracy than the single sensor estimation. This study will provide a reference for the application of proximal soil sensing.

WOVe: Incorporating Word Order in GloVe Word Embeddings

Word vector representations open up new opportunities to extract useful information from unstructured text. Defining a word as a vector made it easy for the machine learning algorithms to understand a text and extract information from. Word vector representations have been used in many applications such word synonyms, word analogy, syntactic parsing, and many others. GloVe, based on word contexts and matrix vectorization, is an effective vector-learning algorithm. It improves on previous vector-learning algorithms. However, the GloVe model fails to explicitly consider the order in which words appear within their contexts. In this paper, multiple methods of incorporating word order in GloVe word embeddings are proposed. Experimental results show that our Word Order Vector (WOVe) word embeddings approach outperforms unmodified GloVe on the natural language tasks of analogy completion and word similarity. WOVe with direct concatenation slightly outperformed GloVe on the word similarity task, increasing average rank by 2%. However, it greatly improved on the GloVe baseline on a word analogy task, achieving an average 36.34% improvement in accuracy.

A Multiscale Attention-Guided UNet++ with Edge Constraint for Building Extraction from High Spatial Resolution Imagery

Building extraction from high spatial resolution imagery (HSRI) plays an important role in the remotely sensed imagery application fields. However, automatically extracting buildings from HSRI is still a challenging task due to such factors as large size variations of buildings, background complexity, variations in appearance, etc. Especially, it is difficult to extract both crowded small buildings and large buildings with accurate boundaries. To address these challenges, this paper presents an end-to-end encoder–decoder model to automatically extract buildings from HSRI. The designed network, called AEUNet++, is based on UNet++, attention mechanism and multi-task learning. Specifically, the AEUNet++ introduces the UNet++ as the backbone to extract multiscale features. Then, the attention block is used to effectively fuse different-layer feature maps instead of direct concatenation in the output of traditional UNet++, which can assign adaptive weights to different-layer feature maps as their relative importance to enhance the sensitivity of the mode and suppress the background influence of irrelevant features. To further improve the boundary accuracy of the extracted buildings, the boundary geometric information of buildings is integrated into the proposed model by a multi-task loss using a proposed distance class map during training of the network, which simultaneously learns the extraction of buildings and boundaries and only outputs extracted buildings while testing. Two different data sets are utilized for evaluating the performance of AEUNet++. The experimental results indicate that AEUNet++ produces greater accuracy than U-Net and the original UNet++ architectures and, hence, provides an effective method for building extraction from HSRI.

PS-Net: Progressive Selection Network for Salient Object Detection

Low-level features contain abundant details and high-level features have rich semantic information. Integrating multi-scale features in an appropriate way is significant for salient object detection. However, direct concatenation or addition taken by most methods ignores the distinctions of contribution among multi-scale features. Besides, most salient object detection models fail to dynamically adjust receptive fields to fit objects of various sizes. To tackle these problems, we propose a Progressive Selection Network (PS-Net). Specifically, PS-Net dynamically extracts high-level features and encourages high-level features to guide low-level features to suppress the background response of the original features. We proposed a salient model PS-Net that selects features progressively at multiply levels. First, we propose a Pyramid Feature Dynamic Extraction module to dynamically select appropriate receptive fields to extract high-level features by Feature Dynamic Extraction modules step by step. Besides, a Self-Interaction Attention module is designed to extract detailed information for low-level features. Finally, we design a Scale Aware Fusion module to fuse these multiple features for adequate exploitation of high-level features to refine low-level features gradually. Compared with 19 start-of-the-art methods on 6 public benchmark datasets, the proposed method achieves remarkable performance in both quantitative and qualitative evaluation. We performed a lot of ablation studies, and more discussions to demonstrate the effectiveness and superiority of our proposed method. In this paper, we propose a PS-Net for effective salient object detection. Extensive experiments on 6 datasets validate that the proposed model outperforms 19 state-of-the-art methods under different evaluation metrics.

TAO-Net: Task-Adaptive Operation Network for Image Restoration and Enhancement

Recently, deep learning based models have been extensively applied to image restoration and enhancement tasks. However, existing approaches neglect the potential correlation among these tasks, and do not fully consider the different intensity factors of degradation which affect image quality. To this end, we propose Task-Adaptive Operation Network (TAO-Net) that stacks a series of operation blocks, guided by the supervised attention mechanism to adaptively enable the model to assign corresponding intensity weights for different degradation factors depending on the input signals. To better recover and preserve the accurate details during the repeated operations, we make full use of the image prior to intruduce structure refinement block as auxiliary information for reconstructing high-quality results. Furthermore, feature aggregation block is also adopted to avoid feature interference caused by the direct concatenation between the backbone operation block and the auxiliary refinement block together. Extensive experiments on multiple benchmark datasets demonstrate that our proposed method outperforms previous baselines in image deraining and low-light image enhancement.

TANet: Target Attention Network for Video Bit-Depth Enhancement

Video bit-depth enhancement (VBDE) reconstructs high-bit-depth (HBD) frames from a low-bit-depth (LBD) video sequence. As neighboring frames contain a considerable amount of complementary information related to the center frame, it is vital for VBDE to exploit neighboring frames as much as possible. Conventional VBDE algorithms with explicit alignment across frames attempt to warp each neighboring frame to the center frame with estimated optical flow, taking into account only pairwise correlation. Most spatiotemporal fusion approaches involve direct concatenation or 3D convolution and treat all features equally, failing to focus on information related to the center frame. Therefore, in this paper, we introduce an improved nonlocal block as a global attentive alignment (GAA) module, which takes the whole input video sequence into consideration to capture features that are globally correlated, to perform implicit alignment. Furthermore, given the bulk of features extracted from the center and neighboring frames, we propose target-guided attention (TGA). TGA can exploit more center-frame-related details and facilitate feature fusion. The proposed network (dubbed TANet) is capable of effectively eliminating false contours and recovering the center frame in high quality, as demonstrated by the experimental results. TANet outperforms state-of-the-art models in terms of both PSNR and SSIM with low time consumption.

Brain-Computer Interface: Applications to Speech Decoding and Synthesis to Augment Communication

Damage or degeneration of motor pathways necessary for speech and other movements, as in brainstem strokes or amyotrophic lateral sclerosis (ALS), can interfere with efficient communication without affecting brain structures responsible for language or cognition. In the worst-case scenario, this can result in the locked in syndrome (LIS), a condition in which individuals cannot initiate communication and can only express themselves by answering yes/no questions with eye blinks or other rudimentary movements. Existing augmentative and alternative communication (AAC) devices that rely on eye tracking can improve the quality of life for people with this condition, but brain-computer interfaces (BCIs) are also increasingly being investigated as AAC devices, particularly when eye tracking is too slow or unreliable. Moreover, with recent and ongoing advances in machine learning and neural recording technologies, BCIs may offer the only means to go beyond cursor control and text generation on a computer, to allow real-time synthesis of speech, which would arguably offer the most efficient and expressive channel for communication. The potential for BCI speech synthesis has only recently been realized because of seminal studies of the neuroanatomical and neurophysiological underpinnings of speech production using intracranial electrocorticographic (ECoG) recordings in patients undergoing epilepsy surgery. These studies have shown that cortical areas responsible for vocalization and articulation are distributed over a large area of ventral sensorimotor cortex, and that it is possible to decode speech and reconstruct its acoustics from ECoG if these areas are recorded with sufficiently dense and comprehensive electrode arrays. In this article, we review these advances, including the latest neural decoding strategies that range from deep learning models to the direct concatenation of speech units. We also discuss state-of-the-art vocoders that are integral in constructing natural-sounding audio waveforms for speech BCIs. Finally, this review outlines some of the challenges ahead in directly synthesizing speech for patients with LIS.

Pan-Sharpening Based on Transformer With Redundancy Reduction

Pan-sharpening methods based on deep neural network (DNN) have produced the state-of-the-art results. However, the common information in the panchromatic (PAN) image and the low spatial resolution multispectral (LRMS) image is not sufficiently explored. As PAN and LRMS images are collected from the same scene, there exists some common information among them, in addition to their respective unique information. The direct concatenation of extracted features leads to some redundancy in the feature space. To reduce the redundancy among features and exploit the global information in source images, we proposed a novel pan-sharpening method by combining the convolution neural network and transformer. Specifically, PAN and LRMS images are encoded as unique features and common features by the subnetworks consisting of convolution blocks and transformer blocks. Then, the common features are averaged and combined with unique features from source images for the reconstruction of the fused image. To extract accurate common features, the equality constraint is imposed on them. Experimental results show that the proposed method outperforms the state-of-the-art methods on both reduced-scale and full-scale datasets. The source code is available at <uri>https://github.com/RSMagneto/TRRNet</uri>.

Neural Reference Synthesis for Inter Frame Coding

This work proposes the neural reference synthesis (NRS) to generate high-fidelity reference block for motion estimation and motion compensation (MEMC) in inter frame coding. The NRS is comprised of two submodules: one for reconstruction enhancement and the other for reference generation. Although numerous methods have been developed in the past for these two submodules using either handcrafted rules or deep convolutional neural network (CNN) models, they basically deal with them separately, resulting in limited coding gains. By contrast, the NRS proposes to optimize them collaboratively. It first develops two CNN-based models, namely EnhNet and GenNet. The EnhNet only uses spatial correlations within the current frame for reconstruction enhancement and the GenNet is then augmented by further aggregating temporal correlations across multiple frames for reference synthesis. However, a direct concatenation of EnhNet and GenNet without considering the complex temporal reference dependency across inter frames would implicitly induce iterative CNN processing and cause the data overfitting problem, leading to visually-disturbing artifacts and oversmoothed pixels. To tackle this problem, the NRS applies a new training strategy to coordinate the EnhNet and GenNet for more robust and generalizable models, and also devises a lightweight multi-level R-D (rate-distortion) selection policy for the encoder to adaptively choose reference blocks generated from the proposed NRS model or conventional coding process. Our NRS not only offers state-of-the-art coding gains, e.g., >10% BD-Rate (Bjøntegaard Delta Rate) reduction against the High Efficiency Video Coding (HEVC) anchor for a variety of common test video sequences encoded at a wide bit range in both low-delay and random access settings, but also greatly reduces the complexity relative to existing learning-based methods by utilizing more lightweight DNNs. All models are made publicly accessible at https://github.com/IVC-Projects/NRS for reproducible research.

Fusion of visible-to-near-infrared and mid-infrared spectroscopy to estimate soil organic carbon

Spectral techniques such as visible-to-near-infrared (VIS–NIR) and mid-infrared (MIR) spectroscopies have been regarded as effective alternatives to laboratory-based methods for determining soil organic carbon (SOC). Research to explore the potential of the fusion of VIS–NIR and MIR absorbance for improving SOC prediction is needed, since each individual spectral range may not contain sufficient information to yield reasonable estimation accuracy. Here, we investigated two data fusion strategies that differed in input data, including direct concatenation of full-spectral absorbance and concatenation of selected predictors by optimal band combination (OBC) algorithm. Specifically, continuous wavelet transform (CWT) was adopted to optimize the spectral data before and after data fusion. Prediction models for SOC were developed using partial least squares regression. Results demonstrated that estimations for SOC using MIR absorbance (i.e., validation R2 = 0.45–0.64) generally outperformed those using VIS–NIR (i.e., validation R2 = 0.20–0.44). Compared to the raw absorbance counterparts, CWT decomposing could improve the prediction accuracy for SOC, for both the individual absorbance and the fusion of VIS–NIR and MIR absorbance. Among all the models investigated, the combinational use of VIS–NIR and MIR using OBC fusion at CWT scale of 1 yielded the optimal prediction, providing the highest validation R2 of 0.66. This model with 10 selected spectral parameters as input is of small total data volume, large processing speed and efficiency, confirming the potential of OBC in fusing both types of spectral data. In summary, CWT decomposing and OBC strategy are powerful algorithms in analyzing the spectral data, and allow the VIS–NIR and MIR spectral fusion models to improve the SOC estimation.

Fused-Deep-Features Based Grape Leaf Disease Diagnosis

Rapid and accurate grape leaf disease diagnosis is of great significance to its yield and quality of grape. In this paper, aiming at the identification of grape leaf diseases, a fast and accurate detection method based on fused deep features, extracted from a convolutional neural network (CNN), plus a support vector machine (SVM) is proposed. In the research, based on an open dataset, three types of state-of-the-art CNN networks, three kinds of deep feature fusion methods, seven species of deep feature layers, and a multi-class SVM classifier were studied. Firstly, images were resized to meet the input requirements of the CNN network; then, the deep features of the input images were extracted via the specific deep feature layer of the CNN network. Two kinds of deep features from different networks were then fused using different fusion methods to increase the effective classification feature information. Finally, a multi-class SVM classifier was trained with the fused deep features. The experimental results on the open dataset show that the fused deep features with any kind of fusion method can obtain a better classification performance than using a single type of deep feature. The direct concatenation of the Fc1000 deep feature extracted from ResNet50 and ResNet101 can achieve the best classification result compared with the other two fusion methods, and its F1 score is 99.81%. Furthermore, the SVM classifier trained using the proposed method can achieve a classification performance comparable to that of using the CNN model directly, but the training time is less than 1 s, which has an advantage over spending tens of minutes training a CNN model. The experimental results indicate that the method proposed in this paper can achieve fast and accurate identification of grape leaf diseases and meet the needs of actual agricultural production.