Multi-Receptive Field Soft Attention Part Learning for Vehicle Re-Identification.

Vehicle re-identification across multiple cameras is one of the main problems of intelligent transportation systems (ITSs). Since the differences in the appearance between different vehicles of the same model are small and the appearance of the same vehicle changes drastically from different viewpoints, vehicle re-identification is a challenging task. In this paper, we propose a model called multi-receptive field soft attention part learning (MRF-SAPL). The MRF-SAPL model learns semantically diverse vehicle part-level features under different receptive fields through multiple local branches, alleviating the problem of small differences in vehicle appearance. To align vehicle parts from different images, this study uses soft attention to adaptively locate the positions of the parts on the final feature map generated by a local branch and maintain the continuity of the internal semantics of the parts. In addition, to obtain parts with different semantic patterns, we propose a new loss function that punishes overlapping regions, forcing the positions of different parts on the same feature map to not overlap each other as much as possible. Extensive ablation experiments demonstrate the effectiveness of our part-level feature learning method MRF-SAPL, and our model achieves state-of-the-art performance on two benchmark datasets.

Breast masses in mammography classification with local contour features.

BACKGROUND: Mammography is one of the most popular tools for early detection of breast cancer. Contour of breast mass in mammography is very important information to distinguish benign and malignant mass. Contour of benign mass is smooth and round or oval, while malignant mass has irregular shape and spiculated contour. Several studies have shown that 1D signature translated from 2D contour can describe the contour features well. METHODS: In this paper, we propose a new method to translate 2D contour of breast mass in mammography into 1D signature. The method can describe not only the contour features but also the regularity of breast mass. Then we segment the whole 1D signature into different subsections. We extract four local features including a new contour descriptor from the subsections. The new contour descriptor is root mean square (RMS) slope. It can describe the roughness of the contour. KNN, SVM and ANN classifier are used to classify benign breast mass and malignant mass. RESULTS: The proposed method is tested on a set with 323 contours including 143 benign masses and 180 malignant ones from digital database of screening mammography (DDSM). The best accuracy of classification is 99.66% using the feature of root mean square slope with SVM classifier. CONCLUSION: The performance of the proposed method is better than traditional method. In addition, RMS slope is an effective feature comparable to most of the existing features.

A framework for retinal vasculature segmentation based on matched filters.

BACKGROUND: Automatic fundus image processing plays a significant role in computer-assisted retinopathy diagnosis. As retinal vasculature is an important anatomical structure in ophthalmic images, recently, retinal vasculature segmentation has received considerable attention from researchers. A segmentation method usually consists of three steps: preprocessing, segmentation, post-processing. Most of the existing methods emphasize on the segmentation step. In our opinion, the vessels and background can be easily separable when suitable preprocessing exists. METHODS: This paper represents a new matched filter-based vasculature segmentation method for 2-D retinal images. First of all, a raw segmentation is acquired by thresholding the images preprocessed using weighted improved circular gabor filter and multi-directional multi-scale second derivation of Gaussian. After that, the raw segmented image is fine-tuned by a set of novel elongating filters. Finally, we eliminate the speckle like regions and isolated pixels, most of which are non-vessel noises and miss-classified fovea or pathological regions. RESULTS: The performance of the proposed method is examined on two popularly used benchmark databases: DRIVE and STARE. The accuracy values are 95.29 and 95.69 %, respectively, without a significant degradation of specificity and sensitivity. CONCLUSION: The performance of the proposed method is significantly better than almost all unsupervised methods, in addition, comparable to most of the existing supervised vasculature segmentation methods.

The Impacts of Cellulose on Volatile Fatty Acid Production and the Microbial Community in Anaerobic Fermentation of Sludge at High and Medium Temperatures.

During large-scale sewage treatment, a large amount of excessive sludge is produced, which will cause serious pollution in the environment. In recent years, anaerobic digestion technology has been widely promoted because it can achieve better sludge reduction, and the products and byproducts after anaerobic digestion can be fully utilized as resources. In this study, cellulose was added as the co-fermentation substrate during the fermentation process at 30 ℃ and 50 ℃ to enhance the production of VFAs. The result indicated that cellulose could significantly increase the yield of VFAs in both 30 ℃ and 50 ℃. Meanwhile, COD and reducing sugar generation in the fermentation process were also measure. Analysis of the microbial community structure at the class and genus levels revealed that the proportion of several genus closely related with cellulose degradation such as Cellvibrio, Fibrobacter, and Sporocytophaga were significantly increased with the addition of cellulose. Co-fermentation was recognized as an economic and environmental friendly strategy for sludge and other solid waste treatment. The analysis of the effect of cellulose as a substrate on the production of VFAs at high and medium temperatures is highly important for exploring ways to increase the production of VFAs in anaerobic fermentation.

Heterogeneous context interaction network for vehicle re-identification.

Capturing global and subtle discriminative information using attention mechanisms is essential to address the challenge of inter-class high similarity for vehicle re-identification (Re-ID) task. Mixing self-information of nodes or modeling context based on pairwise dependencies between nodes are the core ideas of current advanced attention mechanisms. This paper aims to explore how to utilize both dependency context and self-context in an efficient way to facilitate attention to learn more effectively. We propose a heterogeneous context interaction (HCI) attention mechanism that infers the weights of nodes from the interactions of global dependency contexts and local self-contexts to enhance the effect of attention learning. To reduce computational complexity, global dependency contexts are modeled by aggregating number-compressed pairwise dependencies, and the interactions of heterogeneous contexts are restricted to a certain range. Based on this mechanism, we propose a heterogeneous context interaction network (HCI-Net), which uses channel heterogeneous context interaction module (CHCI) and spatial heterogeneous context interaction module (SHCI), and introduces a rigid partitioning strategy to extract important global and fine-grained features. In addition, we design a non-similarity constraint (NSC) that forces the HCI-Net to learn diverse subtle discriminative information. The experiment results on two large datasets, VeRi-776 and VehicleID, show that our proposed HCI-Net achieves the state-of-the-art performance. In particular, the mean average precision (mAP) reaches 83.8% on VeRi-776 dataset.

Predictive Machine Learning Model to Assess the Adsorption Efficiency of Biochar-Heavy Metals for Effective Remediation of Soil-Plant Environment.

Biochar is crucial for agricultural output and plays a significant role in effectively eliminating heavy metals (HMs) from the soil, which is essential for maintaining a soil-plant environment. This work aimed to assess machine learning models to analyze the impact of soil parameters on the transformation of HMs in biochar-soil-plant environments, considering the intricate non-linear relationships involved. A total of 211 datasets from pot or field experiments were evaluated. Fourteen factors were taken into account to assess the efficiency and bioavailability of HM-biochar amendment immobilization. Four predictive models, namely linear regression (LR), partial least squares (PLS), support vector regression (SVR), and random forest (RF), were compared to predict the immobilization efficiency of biochar-HM. The findings revealed that the RF model was created using 5-fold cross-validation, which exhibited a more reliable prediction performance. The results indicated that soil features accounted for 79.7% of the absorption of HM by crops, followed by biochar properties at 17.1% and crop properties at 3.2%. The main elements that influenced the result have been determined as the characteristics of the soil (including the presence of different HM species and the amount of clay) and the quantity and attributes of the biochar (such as the temperature at which it was produced by pyrolysis). Furthermore, the RF model was further developed to predict bioaccumulation factors (BAF) and variations in crop uptake (CCU). The R2 values were found to be 0.7338 and 0.6997, respectively. Thus, machine learning (ML) models could be useful in understanding the behavior of HMs in soil-plant ecosystems by employing biochar additions.

Quantitative Soil Characterization for Biochar-Cd Adsorption: Machine Learning Prediction Models for Cd Transformation and Immobilization.

Soil pollution with cadmium (Cd) poses serious health and environmental consequences. The study investigated the incubation of several soil samples and conducted quantitative soil characterization to assess the influence of biochar (BC) on Cd adsorption. The aim was to develop predictive models for Cd concentrations using statistical and modeling approaches dependent on soil characteristics. The potential risk linked to the transformation and immobilization of Cd adsorption by BC in the soil could be conservatively assessed by pH, clay, cation exchange capacity, organic carbon, and electrical conductivity. In this study, Long Short-Term Memory (LSTM), Bidirectional Gated Recurrent Unit (BiGRU), and 5-layer CNN Convolutional Neural Networks (CNNs) were applied for risk assessments to establish a framework for evaluating Cd risk in BC amended soils to predict Cd transformation. In the case of control soils (CK), the BiGRU model showed commendable performance, with an R2 value of 0.85, indicating an approximate 85.37% variance in the actual Cd. The LSTM model, which incorporates sequence data, produced less accurate results (R2=0.84), while the 5-layer CNN model had an R2 value of 0.91, indicating that the CNN model could account for over 91% of the variation in actual Cd levels. In the case of BC-applied soils, the BiGRU model demonstrated a strong correlation between predicted and actual values with R2 (0.93), indicating that the model explained 93.21% of the variance in Cd concentrations. Similarly, the LSTM model showed a notable increase in performance with BC-treated soil data. The R2 value for this model stands at a robust R2 (0.94), reflecting its enhanced ability to predict Cd levels with BC incorporation. Outperforming both recurrent models, the 5-layer CNN model attained the highest precision with an R2 value of 0.95, suggesting that 95.58% of the variance in the actual Cd data can be explained by the CNN model's predictions in BC-amended soils. Consequently, this study suggests developing ecological soil remediation strategies that can effectively manage heavy metal pollution in soils for environmental sustainability.

Hierarchical level features based trainable segmentation for electron microscopy images.

BACKGROUND: The neuronal electron microscopy images segmentation is the basic and key step to efficiently build the 3D brain structure and connectivity for a better understanding of central neural system. However, due to the visual complex appearance of neuronal structures, it is challenging to automatically segment membranes from the EM images. METHODS: In this paper, we present a fast, efficient segmentation method for neuronal EM images that utilizes hierarchical level features based on supervised learning. Hierarchical level features are designed by combining pixel and superpixel information to describe the EM image. For pixels in a superpixel have similar characteristics, only part of them is automatically selected and used to reduce information redundancy. To each selected pixel, 34 dimensional features are extracted by traditional way. Each superpixel itself is viewed as a unit to extract 35 dimensional features with statistical method. Also, 3 dimensional context level features among multi superpixels are extracted. Above three kinds of features are combined as a feature vector, namely, hierarchical level features to use for segmentation. Random forest is used as classifier and is trained with hierarchical level features to perform segmentation. RESULTS: In small sample condition and with low-dimensional features, the effectiveness of our method is verified on the data set of ISBI2012 EM Segmentation Challenge, and its rand error, warping error and pixel error attain to 0.106308715, 0.001200104 and 0.079132453, respectively. CONCLUSIONS: Comparing to pixel level or superpixel level features, hierarchical level features have better discrimination ability and the proposed method is promising for membrane segmentation.

Video-based fingerprint verification.

Conventional fingerprint verification systems use only static information. In this paper, fingerprint videos, which contain dynamic information, are utilized for verification. Fingerprint videos are acquired by the same capture device that acquires conventional fingerprint images, and the user experience of providing a fingerprint video is the same as that of providing a single impression. After preprocessing and aligning processes, "inside similarity" and "outside similarity" are defined and calculated to take advantage of both dynamic and static information contained in fingerprint videos. Match scores between two matching fingerprint videos are then calculated by combining the two kinds of similarity. Experimental results show that the proposed video-based method leads to a relative reduction of 60 percent in the equal error rate (EER) in comparison to the conventional single impression-based method. We also analyze the time complexity of our method when different combinations of strategies are used. Our method still outperforms the conventional method, even if both methods have the same time complexity. Finally, experimental results demonstrate that the proposed video-based method can lead to better accuracy than the multiple impressions fusion method, and the proposed method has a much lower false acceptance rate (FAR) when the false rejection rate (FRR) is quite low.

Sliding window-based region of interest extraction for finger vein images.

Region of Interest (ROI) extraction is a crucial step in an automatic finger vein recognition system. The aim of ROI extraction is to decide which part of the image is suitable for finger vein feature extraction. This paper proposes a finger vein ROI extraction method which is robust to finger displacement and rotation. First, we determine the middle line of the finger, which will be used to correct the image skew. Then, a sliding window is used to detect the phalangeal joints and further to ascertain the height of ROI. Last, for the corrective image with certain height, we will obtain the ROI by using the internal tangents of finger edges as the left and right boundary. The experimental results show that the proposed method can extract ROI more accurately and effectively compared with other methods, and thus improve the performance of finger vein identification system. Besides, to acquire the high quality finger vein image during the capture process, we propose eight criteria for finger vein capture from different aspects and these criteria should be helpful to some extent for finger vein capture.

Retinal identification based on an Improved Circular Gabor Filter and Scale Invariant Feature Transform.

Retinal identification based on retinal vasculatures in the retina provides the most secure and accurate means of authentication among biometrics and has primarily been used in combination with access control systems at high security facilities. Recently, there has been much interest in retina identification. As digital retina images always suffer from deformations, the Scale Invariant Feature Transform (SIFT), which is known for its distinctiveness and invariance for scale and rotation, has been introduced to retinal based identification. However, some shortcomings like the difficulty of feature extraction and mismatching exist in SIFT-based identification. To solve these problems, a novel preprocessing method based on the Improved Circular Gabor Transform (ICGF) is proposed. After further processing by the iterated spatial anisotropic smooth method, the number of uninformative SIFT keypoints is decreased dramatically. Tested on the VARIA and eight simulated retina databases combining rotation and scaling, the developed method presents promising results and shows robustness to rotations and scale changes.

Finger vein recognition with personalized feature selection.

Finger veins are a promising biometric pattern for personalized identification in terms of their advantages over existing biometrics. Based on the spatial pyramid representation and the combination of more effective information such as gray, texture and shape, this paper proposes a simple but powerful feature, called Pyramid Histograms of Gray, Texture and Orientation Gradients (PHGTOG). For a finger vein image, PHGTOG can reflect the global spatial layout and local details of gray, texture and shape. To further improve the recognition performance and reduce the computational complexity, we select a personalized subset of features from PHGTOG for each subject by using the sparse weight vector, which is trained by using LASSO and called PFS-PHGTOG. We conduct extensive experiments to demonstrate the promise of the PHGTOG and PFS-PHGTOG, experimental results on our databases show that PHGTOG outperforms the other existing features. Moreover, PFS-PHGTOG can further boost the performance in comparison with PHGTOG.

Finger vein recognition based on personalized weight maps.

Finger vein recognition is a promising biometric recognition technology, which verifies identities via the vein patterns in the fingers. Binary pattern based methods were thoroughly studied in order to cope with the difficulties of extracting the blood vessel network. However, current binary pattern based finger vein matching methods treat every bit of feature codes derived from different image of various individuals as equally important and assign the same weight value to them. In this paper, we propose a finger vein recognition method based on personalized weight maps (PWMs). The different bits have different weight values according to their stabilities in a certain number of training samples from an individual. Firstly we present the concept of PWM, and then propose the finger vein recognition framework, which mainly consists of preprocessing, feature extraction, and matching. Finally, we design extensive experiments to evaluate the effectiveness of our proposal. Experimental results show that PWM achieves not only better performance, but also high robustness and reliability. In addition, PWM can be used as a general framework for binary pattern based recognition.

[Ultrafiltration membrane extract mixture from Angelica sinensis and Hedysarum polybotrys induced transdifferentiation of BMSCs in mice: an experimental research].

OBJECTIVE: To observe and evaluate the effect of transdifferentiation of bone marrow derived stroma cells (BMSCs) into nerve cells by ultrafiltration membrane extract mixture from Angelica sinensis and Hedysarum polybotrys. METHODS: The BMSCs in vitro cultured after treated by ultrafiltration membrane extract mixture from Angelica sinensis and Hedysarum polybotrys were divided into 5 groups, i.e., the blank group, the low dose group (6 g/L mixture), the high dose group (12 g/L mixture), the combination group (3 g/L mixture + 0.5 mmol/Lbeta-mercaptoethanol), and the positive control group (13-mercaptoethanol). The effects of transdifferentiation of nerve cells were observed using toluidine blue staining in each group. The differences of 5 specific neuroproteins, i.e. neuron-specific enolase (NSE), nestin, neurofilament protein (NFP), microtubule associated protein 2 (MAP2), and glial fibrillary acidic protein (GFAP) were detected using immunohistochemical technique and immunofluorescent technique respectively. The changes of the cell cycle were detected using flow cytometry (FCM). RESULTS: After induction BMSCs changed morphologically. The morphological features were weaker in the high and low dose groups than in the combination group and the positive group. Except the blank group, the aforesaid 5 proteins expressed positively in the rest groups. Their expression levels were highest in the positive control group (P <0.05), followed by the combination group (P <0.05). As for the cell proliferation rate detected by FCM, it was the lowest in the positive control group, followed by high dose group, low dose group, and then the combination group (all P <0.05). CONCLUSIONS: The ultrafiltration membrane extract mixture from Angelica sinensis and Hedysarum polybotrys could effectively induce the transdifferentiation of BMSCs into nerve cells. Its inducing capacities were weaker in the positive control group, but it showed marked proliferation effects on differentiated cells. Therefore, the mixture might be a more ideal medication pathway for effectively inducing BMSCs' transdifferentiation into nerve cells, which might have higher proliferation and be used for clinical research.

Towards Accurate and Robust Domain Adaptation Under Multiple Noisy Environments.

In many non-stationary environments, machine learning algorithms usually confront the distribution shift scenarios. Previous domain adaptation methods have achieved great success. However, they would lose algorithm robustness in multiple noisy environments where the examples of source domain become corrupted by label noise, feature noise, or open-set noise. In this paper, we report our attempt toward achieving noise-robust domain adaptation. We first give a theoretical analysis and find that different noises have disparate impacts on the expected target risk. To eliminate the effect of source noises, we propose offline curriculum learning minimizing a newly-defined empirical source risk. We suggest a proxy distribution-based margin discrepancy to gradually decrease the noisy distribution distance to reduce the impact of source noises. We propose an energy estimator for assessing the outlier degree of open-set-noise examples to defeat the harmful influence. We also suggest robust parameter learning to mitigate the negative effect further and learn domain-invariant feature representations. Finally, we seamlessly transform these components into an adversarial network that performs efficient joint optimization for them. A series of empirical studies on the benchmark datasets and the COVID-19 screening task show that our algorithm remarkably outperforms the state-of-the-art, with over 10% accuracy improvements in some transfer tasks.

Reformulating Graph Kernels for Self-Supervised Space-Time Correspondence Learning.

Self-supervised space-time correspondence learning utilizing unlabeled videos holds great potential in computer vision. Most existing methods rely on contrastive learning with mining negative samples or adapting reconstruction from the image domain, which requires dense affinity across multiple frames or optical flow constraints. Moreover, video correspondence prediction models need to uncover more inherent properties of the video, such as structural information. In this work, we propose HiGraph+, a sophisticated space-time correspondence framework based on learnable graph kernels. By treating videos as a spatial-temporal graph, the learning objective of HiGraph+ is issued in a self-supervised manner, predicting the unobserved hidden graph via graph kernel methods. First, we learn the structural consistency of sub-graphs in graph-level correspondence learning. Furthermore, we introduce a spatio-temporal hidden graph loss through contrastive learning that facilitates learning temporal coherence across frames of sub-graphs and spatial diversity within the same frame. Therefore, we can predict long-term correspondences and drive the hidden graph to acquire distinct local structural representations. Then, we learn a refined representation across frames on the node-level via a dense graph kernel. The structural and temporal consistency of the graph forms the self-supervision of model training. HiGraph+ achieves excellent performance and demonstrates robustness in benchmark tests involving object, semantic part, keypoint, and instance labeling propagation tasks. Our algorithm implementations have been made publicly available at https://github.com/zyqin19/HiGraph.

Missingness-Pattern-Adaptive Learning With Incomplete Data.

Many real-world problems deal with collections of data with missing values, e.g., RNA sequential analytics, image completion, video processing, etc. Usually, such missing data is a serious impediment to a good learning achievement. Existing methods tend to use a universal model for all incomplete data, resulting in a suboptimal model for each missingness pattern. In this paper, we present a general model for learning with incomplete data. The proposed model can be appropriately adjusted with different missingness patterns, alleviating competitions between data. Our model is based on observable features only, so it does not incur errors from data imputation. We further introduce a low-rank constraint to promote the generalization ability of our model. Analysis of the generalization error justifies our idea theoretically. In additional, a subgradient method is proposed to optimize our model with a proven convergence rate. Experiments on different types of data show that our method compares favorably with typical imputation strategies and other state-of-the-art models for incomplete data. More importantly, our method can be seamlessly incorporated into the neural networks with the best results achieved. The source code is released at https://github.com/YS-GONG/missingness-patterns.

Biomarkers-Aware Asymmetric Bibranch GAN with Adaptive Memory Batch Normalization for Prediction of Anti-VEGF Treatment Response in Neovascular Age-related Macular Degeneration.

The emergence of anti-vascular endothelial growth factor (anti-VEGF) therapy has revolutionized neovascular age-related macular degeneration (nAMD). Post-therapeutic optical coherence tomography (OCT) imaging facilitates the prediction of therapeutic response to anti-VEGF therapy for nAMD. Although the generative adversarial network (GAN) is a popular generative model for post-therapeutic OCT image generation, it is realistically challenging to gather sufficient pre- and post-therapeutic OCT image pairs, resulting in overfitting. Moreover, the available GAN-based methods ignore local details, such as the biomarkers that are essential for nAMD treatment. To address these issues, a Biomarkers-aware Asymmetric Bibranch GAN (BAABGAN) is proposed to efficiently generate post-therapeutic OCT images. Specifically, one branch is developed to learn prior knowledge with a high degree of transferability from large-scale data, termed the source branch. Then, the source branch transfer knowledge to another branch, which is trained on small-scale paired data, termed the target branch. To boost the transferability, a novel Adaptive Memory Batch Normalization (AMBN) is introduced in the source branch, which learns more effective global knowledge that is impervious to noise via memory mechanism. Also, a novel Adaptive Biomarkers-aware Attention (ABA) module is proposed to encode biomarkers information into latent features of target branches to learn finer local details of biomarkers. The proposed method outperforms traditional GAN models and can produce high-quality post-treatment OCT pictures with limited data sets, as shown by the results of experiments.

Finger vein recognition based on (2D)² PCA and metric learning.

Finger vein recognition is a promising biometric recognition technology, which verifies identities via the vein patterns in the fingers. In this paper, (2D)² PCA is applied to extract features of finger veins, based on which a new recognition method is proposed in conjunction with metric learning. It learns a KNN classifier for each individual, which is different from the traditional methods where a fixed threshold is employed for all individuals. Besides, the SMOTE technology is adopted to solve the class-imbalance problem. Our experiments show that the proposed method is effective by achieving a recognition rate of 99.17%.

Two-level evaluation on sensor interoperability of features in fingerprint image segmentation.

Features used in fingerprint segmentation significantly affect the segmentation performance. Various features exhibit different discriminating abilities on fingerprint images derived from different sensors. One feature which has better discriminating ability on images derived from a certain sensor may not adapt to segment images derived from other sensors. This degrades the segmentation performance. This paper empirically analyzes the sensor interoperability problem of segmentation feature, which refers to the feature's ability to adapt to the raw fingerprints captured by different sensors. To address this issue, this paper presents a two-level feature evaluation method, including the first level feature evaluation based on segmentation error rate and the second level feature evaluation based on decision tree. The proposed method is performed on a number of fingerprint databases which are obtained from various sensors. Experimental results show that the proposed method can effectively evaluate the sensor interoperability of features, and the features with good evaluation results acquire better segmentation accuracies of images originating from different sensors.