Comparative genomics analysis provide insights into evolution and stress responses of Lhcb genes in Rosaceae fruit crops.

BACKGROUND: Light-harvesting chlorophyll a/b b evelopment of higher plants and in response to abiotic stress. Previous works has demonstrated that that Lhcb genes were involved in the phytochrome regulation and responded to the different light and temperature conditions in Poaceae (such as maize). However, the evolution and functions of Lhcb genes remains poorly characterized in important Rosaceae species. RESULTS: In this investigation, we conducted a genome-wide analysis and identified a total of 212 Lhcb genes across nine Rosaceae species. Specifically, we found 23 Lhcb genes in Fragaria vesca, 20 in Prunus armeniaca, 33 in Malus domestica 'Gala', 21 in Prunus persica, 33 in Rosa chinensis, 29 in Pyrus bretschneideri, 18 in Rubus occidentalis, 20 in Prunus mume, and 15 in Prunus salicina. Phylogenetic analysis revealed that the Lhcb gene family could be classified into seven major subfamilies, with members of each subfamily sharing similar conserved motifs. And, the functions of each subfamily was predicted based on the previous reports from other species. The Lhcb proteins were highly conserved within their respective subfamilies, suggesting similar functions. Interestingly, we observed similar peaks in Ks values (0.1-0.2) for Lhcb genes in apple and pear, indicating a recent whole genome duplication event (about 30 to 45 million years ago). Additionally, a few Lhcb genes underwent tandem duplication and were located across all chromosomes of nine species of Rosaceae. Furthermore, the analysis of the cis-acting elements in the 2000 bp promoter region upstream of the pear Lhcb gene revealed four main categories: light response correlation, stress response correlation, hormone response correlation, and plant growth. Quantitative expression analysis demonstrated that Lhcb genes exhibited tissue-specific expression patterns and responded differently to low-temperature stress in Rosaceae species. CONCLUSIONS: These findings shed light on the evolution and phylogeny of Lhcb genes in Rosaceae and highlight the critical role of Lhcb in pear's response to low temperatures. The results obtained provide valuable insights for further investigations into the functions of Lhcb genes in Rosaceae, and these functional genes will be used for further fruit tree breeding and improvement to cope with the current climate changes.

Diagnosing the benign paroxysmal positional vertigo via 1D and deep-learning composite model.

BACKGROUND: Benign Paroxysmal Positional Vertigo (BPPV) is the leading cause of vertigo, and its characteristic nystagmus induced by positional maneuvers makes it a good model for Artificial Intelligence (AI) diagnosis. However, during the testing procedure, up to 10 min of indivisible long-range temporal correlation data are produced, making the AI-informed real-time diagnosing unlikely in clinical practice. METHODS: A combined 1D and Deep-Learning (DL) composite model was proposed. Two separate cohorts were recruited, with one for model generation and the other for evaluation of model's real-world generalizability. Eight features, including two head traces and three eye traces and their corresponding slow phase velocity (SPV) value, were served as the inputs. Three candidate models were tested, and a sensitivity study was conducted to determine the saliently important features. RESULTS: The study included 2671 patients in the training cohort and 703 in the test cohort. A hybrid DL model achieved a micro-area under the receiver operating curve (AUROC) of 0.982 (95% CI 0.965, 0.994) and macro-AUROC of 0.965 (95% CI 0.898, 0.999) for overall classification. The highest accuracy was observed for right posterior BPPV, with an AUROC of 0.991 (95% CI 0.972, 1.000), followed by left posterior BPPV, with an AUROC of 0.979 (95% CI 0.940, 0.998), the lowest AUROC was 0.928 (95% CI 0.878, 0.966) for lateral BPPV. The SPV was consistently identified as the most predictive feature in the models. If the model process is carried out 100 times for a 10-min data, one single running takes 0.79 ± 0.06 s. CONCLUSION: This study designed DL models which can accurately detect and categorize the subtype of BPPV, enabling a quick and straightforward diagnosis of BPPV in clinical setting. The critical feature identified in the model helps expand our understanding of this disorder.

Unsupervised Monocular Visual Odometry for Fast-Moving Scenes Based on Optical Flow Network with Feature Point Matching Constraint.

Robust and accurate visual feature tracking is essential for good pose estimation in visual odometry. However, in fast-moving scenes, feature point extraction and matching are unstable because of blurred images and large image disparity. In this paper, we propose an unsupervised monocular visual odometry framework based on a fusion of features extracted from two sources, that is, the optical flow network and the traditional point feature extractor. In the training process, point features are generated for scene images and the outliers of matched point pairs are filtered by FlannMatch. Meanwhile, the optical flow network constrained by the principle of forward-backward flow consistency is used to select another group of corresponding point pairs. The Euclidean distance between the matching points found by FlannMatch and the corresponding point pairs by the flow network is added to the loss function of the flow network. Compared with SURF, the trained flow network shows more robust performance in complicated fast-motion scenarios. Furthermore, we propose the AvgFlow estimation module, which selects one group of the matched point pairs generated by the two methods according to the scene motion. The camera pose is then recovered by Perspective-n-Point (PnP) or the epipolar geometry. Experiments conducted on the KITTI Odometry dataset verify the effectiveness of the trajectory estimation of our approach, especially in fast-moving scenarios.

Exogenous treatment with melatonin enhances waterlogging tolerance of kiwifruit plants.

Waterlogging stress has an enormous negative impact on the kiwifruit yield and quality. The protective role of exogenous melatonin on water stress has been widely studied, especially in drought stress. However, the research on melatonin-induced waterlogging tolerance is scarce. Here, we found that treatment with exogenous melatonin could effectively alleviate the damage on kiwifruit plants in response to waterlogging treatment. This was accompanied by higher antioxidant activity and lower ROS accumulation in kiwifruit roots during stress period. The detection of changes in amino acid levels of kiwifruit roots during waterlogging stress showed a possible interaction between melatonin and amino acid metabolism, which promoted the tolerance of kiwifruit plants to waterlogging. The higher levels of GABA and Pro in the roots of melatonin-treated kiwifruit plants partly contributed to their improved waterlogging tolerance. In addition, some plant hormones were also involved in the melatonin-mediated waterlogging tolerance, such as the enhancement of ACC accumulation. This study discussed the melatonin-mediated water stress tolerance of plants from the perspective of amino acid metabolism for the first time.

Conuping BSA-Seq and RNA-Seq Reveal the Molecular Pathway and Genes Associated with the Plant Height of Foxtail Millet (Setaria italica).

Foxtail millet (Setaria italica) plays an important role in C4 crop research and agricultural development in arid areas due to its short growth period, drought tolerance, and barren tolerance. Exploration of the dwarfing mechanism and the dwarf genes of foxtail millet can provide a reference for dwarf breeding and dwarf research of other C4 crops. In this study, genetic analysis was performed using phenotypic data, candidate genes were screened by bulk segregant analysis sequencing (BSA-Seq); differentially expressed genes and metabolic pathways in different strains of high samples were analyzed by RNA sequencing (RNA-Seq). The association analysis of BSA-Seq and RNA-Seq further narrowed the candidate range. As a result, a total of three quantitative trait loci (QTLs) and nine candidate genes related to plant height were obtained on chromosomes I and IX. Based on the functional prediction of the candidate genes, we propose a hypothetical mechanism for the formation of millet dwarfing, in which, metabolism and MAPK signaling play important roles in the formation of foxtail millet plant height.

Comparative transcriptome provides insight into responding mechanism of waterlogging stress in Actinidia valvata Dunn.

Kiwifruit is one of the most popular fruits, and the area of its cultivation in China has grown rapidly over the last decade. However, kiwifruit vines are vulnerable to waterlogging, especially in the extensive areas of south China where it is grown. This has become an important factor limiting yields. Therefore, it is necessary to clarify the responses of kiwifruit to waterlogging. Here, we have selected Actinidia valvata Dunn which is able to withstand waterlogging conditions and the waterlogging-susceptible Actinidia deliciosa to perform the RNA-seq of roots under waterlogging stress. Seedling roots of Actinidia valvata Dunn and Actinidia deliciosa presented distinct root phenotypes after waterlogging treatments. Genome mapping showed a large genome difference between Actinidia valvata Dunn and Actinidia deliciosa. Transcription factors MYB, MYB-related, AP2-EREBP, bHLH, WRKY, and NAC were identified as the key genes involved in the response to waterlogging stress of kiwifruit. Meanwhile, the MAPK signaling pathway and the glycolysis/gluconeogenesis pathway were identified as the vital pathways involved in the response to waterlogging, and key genes were identified from these two pathways. These results will broaden our understanding of transcriptional response of waterlogging stress and will provide new insights into the molecular mechanisms associated with waterlogging stress. Furthermore, identification of the genes responsible will assist in the breeding of kiwifruit tolerant of waterlogging.

Conjunctive Analyses of Bulk Segregant Analysis Sequencing and Bulk Segregant RNA Sequencing to Identify Candidate Genes Controlling Spikelet Sterility of Foxtail Millet.

Foxtail millet has gradually become a model gramineous C4 crop owing to its short growth period and small genome. Research on the development of its spikelets is not only directly related to the yield and economic value of foxtail millet but also can provide a reference for studying the fertility of other C4 crops. In this study, a hybrid population containing 200 offspring was constructed from the Xinong8852 and An15 parental lines, and two extreme trait populations were constructed from the F2 generation for analysis of the spikelet sterility. The F2 population conformed to a 3:1 Mendelian segregation ratio, and it was thus concluded that this trait is likely controlled by a single recessive gene. Bulk segregant analysis sequencing (BSA-Seq) was used to determine the candidate regions and candidate genes related to the development of foxtail millet spikelets. Additionally, the functional analysis of differentially expressed genes in populations with different traits was conducted by bulk segregant RNA sequencing (BSR-Seq). Finally, conjunctive analysis of BSA-Seq and BSR-Seq results, combined with biological information analysis, revealed six genes on chromosome VII that were ultimately identified as candidate genes controlling foxtail millet spikelet development. This study provides a new reference for research on foxtail millet sterility and lays a solid foundation for the examination of fertility in other gramineous crops.

W-Transformer: Accurate Cobb angles estimation by using a transformer-based hybrid structure.

BACKGROUND: Scoliosis is a type of spinal deformity, which is harmful to a person's health. In severe cases, it can trigger paralysis or death. The measurement of Cobb angle plays an essential role in assessing the severity of scoliosis. PURPOSE: The aim of this paper is to propose an automatic system for landmark detection and Cobb angle estimation, which can effectively help clinicians diagnose and treat scoliosis. METHODS: A novel hybrid framework was proposed to measure Cobb angle precisely for clinical diagnosis, which was referred as W-Transformer due to its w-shaped architecture. First, a convolutional neural network of cascade residual blocks as our backbone was designed. Then a transformer was fused to learn the dependency information between spine and landmarks. In addition, a reinforcement branch was designed to improve the overlap of landmarks, and an improved prediction module was proposed to fine-tune the final coordinates of landmarks in Cobb angles estimation. Besides, the public Accurate Automated Spinal Curvature Estimation (AASCE) MICCAI 2019 challenge was served as data set. It supplies 609 manually labeled spine anterior-posterior (AP) X-ray images, each of which contains a total of 68 landmark labels and three Cobb Angles tags. RESULTS: From the perspective of the AASCE MICCAI 2019 challenge, we achieved a lower symmetric mean absolute percentage error (SMAPE) of 8.26% for all Cobb angles and the lowest averaged detection error of 50.89 in terms of landmark detection, compared with many state-of-the-art methods. We also provided the SMAPEs for the Cobb angles of the proximal-thoracic (PT), the main-thoracic (MT), and the thoracic-lumbar (TL) area, which are 5.27%, 14.59%, and 20.97% respectively, however, these data were not covered in most previous studies. Statistical analysis demonstrates that our model has obtained a high level of Pearson correlation coefficient of 0.9398 ( p < 0.001 $p<0.001$ ), which shows excellent reliability of our model. Our model can yield 0.9489 ( p < 0.001 $p<0.001$ ), 0.8817 ( p < 0.001 $p<0.001$ ), and 0.9149 ( p < 0.001 $p<0.001$ ) for PT, MT, and TL, respectively. The overall variability of Cobb angle measurement is less than 4 ∘ $^\circ$ , implying clinical value. And the mean absolute deviation (standard deviation) for three regions is 3.64 ∘ $^\circ$ (4.13 ∘ $^\circ$ ), 3.84 ∘ $^\circ$ (4.66 ∘ $^\circ$ ), and 3.80 ∘ $^\circ$ (4.19 ∘ $^\circ$ ). The results of Student paired t $t$ -test indicate that no statistically significant differences are observed between manual measurement and our automatic approach ( p $p$ -value is always > $>$ 0.05). Regarding the diagnosis of scoliosis (Cobb angle > $>$ 10 ∘ $^\circ$ ), the proposed method achieves a high sensitivity of 0.9577 and a specificity of 0.8475 for all spinal regions. CONCLUSIONS: This study offers a brand-new automatic approach that is potentially of great benefit of the complex task of landmark detection and Cobb angle evaluation, which can provide helpful navigation information about the early diagnosis of scoliosis.

The analysis of isolation measures for epidemic control of COVID-19.

This paper proposes a susceptible exposed infectious recovered model (SEIR) with isolation measures to evaluate the COVID-19 epidemic based on the prevention and control policy implemented by the Chinese government on February 23, 2020. According to the Chinese government's immediate isolation and centralized diagnosis of confirmed cases, and the adoption of epidemic tracking measures on patients to prevent further spread of the epidemic, we divide the population into susceptible, exposed, infectious, quarantine, confirmed and recovered. This paper proposes an SEIR model with isolation measures that simultaneously investigates the infectivity of the incubation period, reflects prevention and control measures and calculates the basic reproduction number of the model. According to the data released by the National Health Commission of the People's Republic of China, we estimated the parameters of the model and compared the simulation results of the model with actual data. We have considered the trend of the epidemic under different incubation periods of infectious capacity. When the incubation period is not contagious, the peak number of confirmed in the model is 33,870; and when the infectious capacity is 0.1 times the infectious capacity in the infectious period, the peak number of confirmed in the model is 57,950; when the infectious capacity is doubled, the peak number of confirmed will reach 109,300. Moreover, by changing the contact rate in the model, we found that as the intensity of prevention and control measures increase, the peak of the epidemic will come earlier, and the peak number of confirmed will also be significantly reduced. Under extremely strict prevention and control measures, the peak number of confirmed cases has dropped by nearly 50%. In addition, we use the EEMD method to decompose the time series data of the epidemic, and then combine the LSTM model to predict the trend of the epidemic. Compared with the method of directly using LSTM for prediction, more detailed information can be obtained.

Deep learning network for medical volume data segmentation based on multi axial plane fusion.

BACKGROUND AND OBJECTIVE: High-dimensional data generally contains more accurate information for medical image, e.g., computerized tomography (CT) data can depict the three dimensional structure of organs more precisely. However, the data in high-dimension often needs enormous computation and has high memory requirements in the deep learning convolution networks, while dimensional reduction usually leads to performance degradation. METHODS: In this paper, a two-dimensional deep learning segmentation network was proposed for medical volume data based on multi-pinacoidal plane fusion to cover more information under the control of computation.This approach has conducive compatibility while using the model proposed to extract the global information between different inputs layers. RESULTS: Our approach has worked in different backbone network. Using the approach, DeepUnet's Dice coefficient (Dice) and Positive Predictive Value (PPV) are 0.883 and 0.982 showing the satisfied progress. Various backbones can enjoy the profit of the method. CONCLUSIONS: Through the comparison of different backbones, it can be found that the proposed network with multi-pinacoidal plane fusion can achieve better results both quantitively and qualitatively.

Proso millet (Panicum miliaceum L.): A potential crop to meet demand scenario for sustainable saline agriculture.

Proso millet (Panicum miliaceum L.) is resilient to abiotic stress, especially to land degradation caused by soil salinization. However, the mechanisms by which its roots adapt and tolerate salt stress are obscure. In this study, plants of a salt-sensitive cultivar (SS 212) and a salt-tolerant cultivar (ST 47) of proso millet were exposed to severe salt stress and subsequent re-watering. ST 47 exhibited greater salt tolerance than SS 212, as evidenced by higher increases in total root length (TRL), root surface area (RSA), root tip number (RTN). Moreover, microstructural analysis showed that relative to SS 212, the roots of ST 47 could maintain more intact internal structures and thicker cell walls under salt stress. Digital RNA sequence analysis revealed that ST 47 maintained better Na+/K+ balance to resist Na+ toxicity via a higher capability to restrict Na+ uptake, vacuolar Na+ sequestration, and Na+ exclusion. The mechanism for Na+ toxicity resistance in ST 47 involved promoting cell wall composition changes via efficient regulation of galactose metabolism and biosynthesis of cellulose and phenylpropanoids. Overall, this study provides valuable salt-tolerant cultivar resources and mechanisms for regulating salt tolerance, which could be applied for the rehabilitation of saline lands.

Safety and efficacy of a nerve matrix membrane as a collagen nerve wrapping: a randomized, single-blind, multicenter clinical trial.

A new nerve matrix membrane derived from decellularized porcine nerves has been shown to retain the major extracellular matrix components, and to be effective in preventing adhesion between the nerve anastomosis sites and the surrounding tissues in a rat sciatic nerve transection model, thereby enhancing regeneration of the nerve. The effectiveness of the membrane may be attributed to its various bioactive components. In this prospective, randomized, single-blind, parallel-controlled multicenter clinical trial, we compared the safety and efficacy of the new nerve matrix membrane with a previously approved bovine tendon-derived type I collagen nerve wrapping. A total of 120 patients with peripheral nerve injury were recruited from Beijing Jishuitan Hospital, The First Bethune Hospital of Jilin University, and Yantai Yuhuangding Hospital, China. The patients were randomly assigned to undergo end-to-end and tension-free neurorrhaphy with nerve matrix membrane (n = 60, 52 male, 8 female, mean age 41.34 years, experimental group) or tendon-derived collagen nerve wrapping (n = 60, 42 male, 18 female, mean age 40.17 years, control group). Patients were followed-up at 14 ± 5, 30 ± 7, 90 ± 10 and 180 ± 20 days after the operation. Safety evaluation included analyses of local and systemic reactions, related laboratory tests, and adverse reactions. Efficacy evaluation included a static 2-point discrimination test, a moving 2-point discrimination test, and a Semmes-Weinstein monofilament examination. Sensory nerve function was evaluated with the British Medical Research Council Scale and Semmes-Weinstein monofilament examination. The ratio (percentage) of patients with excellent to good results in sensory nerve recovery 180 ± 20 days after the treatment was used as the primary effectiveness index. The percentages of patients with excellent to good results in the experimental and control groups were 98.00% and 94.44%, respectively, with no significant difference between the two groups. There were no significant differences in the results of routine blood tests, liver and renal function tests, coagulation function tests, or immunoglobulin tests at 14 and 180 days postoperatively between the two groups. These findings suggest that the novel nerve matrix membrane is similar in efficacy to the commercially-available bovine-derived collagen membrane in the repair of peripheral nerve injury, and it may therefore serve as an alternative in the clinical setting. The clinical trial was approved by the Institutional Ethics Committee of Beijing Jishuitan Hospital, China (approval No. 20160902) on October 8, 2016, the Institutional Ethics Committee of the First Bethune Hospital of Jilin University, China (approval No. 160518-088) on December 14, 2016, and the Institutional Ethics Committee of Yantai Yuhuangding Hospital, China (approval No. 2016-10-01) on December 9, 2016. The clinical trial was registered with the Chinese Clinical Trial Registry (registration number: ChiCTR2000033324) on May 28, 2020.

Segmentation of coronary arteries images using global feature embedded network with active contour loss.

Coronary heart disease (CHD) is a serious disease that endangers human health and life. In recent years, the morbidity and mortality of CHD are increasing significantly. Because of the particularity and complexity of medical image, it is challenging to segment coronary artery accurately and efficiently. This paper proposes a novel global feature embedded network for better coronary arteries segmentation in 3D coronary computed tomography angiography (CTA) data. The global feature combines multi-level layers from various stages of the network, which contains semantic information and detailed features, aiming to accurately segment target with precise boundary. In addition, we integrate a group of improved noisy activating functions with parameters into our network to eliminate the impact of noise in CTA data. And we improve the learning active contour model, which obtains a refined segmentation result with smooth boundary based on the high-quality score map produced by the networks. The experimental results show that the proposed framework achieved the state-of-the-art performance intuitively and quantitively.

ADR-Net: Context extraction network based on M-Net for medical image segmentation.

PURPOSE: Medical image segmentation is an essential component of medical image analysis. Accurate segmentation can assist doctors in diagnosis and relieve their fatigue. Although several image segmentation methods based on U-Net have been proposed, their performances have been observed to be suboptimal in the case of small-sized objects. To address this shortcoming, a novel network architecture is proposed in this study to enhance segmentation performance on small medical targets. METHODS: In this paper, we propose a joint multi-scale context attention network architecture to simultaneously capture higher level semantic information and spatial information. In order to obtain a greater number of feature maps during decoding, the network concatenates the images of side inputs by down-sampling during the encoding phase. In the bottleneck layer of the network, dense atrous convolution (DAC) and multi-scale residual pyramid pooling (RMP) modules are exploited to better capture high-level semantic information and spatial information. To improve the segmentation performance on small targets, the attention gate (AG) block is used to effectively suppress feature activation in uncorrelated regions and highlight the target area. RESULTS: The proposed model is first evaluated on the public dataset DRIVE, on which it performs significantly better than the basic framework in terms of sensitivity (SE), intersection-over-union (IOU), and area under the receiver operating characteristic curve (AUC). In particular, the SE and IOU are observed to increase by 7.46% and 5.97%, respectively. Further, the evaluation indices exhibit improvements compared to those of state-of-the-art methods as well, with SE and IOU increasing by 3.58% and 3.26%, respectively. Additionally, in order to demonstrate the generalizability of the proposed architecture, we evaluate our model on three other challenging datasets. The respective performances are observed to be better than those of state-of-the-art network architectures on the same datasets. Moreover, we use lung segmentation as a comparative experiment to demonstrate the transferability of the advantageous properties of the proposed approach in the context of small target segmentation to the segmentation of large targets. Finally, an ablation study is conducted to investigate the individual contributions of the AG block, the DAC block, and the RMP block to the performance of the network. CONCLUSIONS: The proposed method is evaluated on various datasets. Experimental results demonstrate that the proposed model performs better than state-of-the-art methods in medical image segmentation of small targets.

LncRNA EWSAT1 upregulates CPEB4 via miR-330-5p to promote cervical cancer development.

Long non-coding RNA (lncRNA) Ewing sarcoma associated transcript 1 (EWSAT1) is an oncogene in a variety of tumors. Here, we planned to demonstrate EWSAT1 function in cervical cancer and further illustrate its underlying mechanism. EWSAT1 expression in cervical cancer was evaluated through qRT-PCR. Colony forming capacity was measured by colony formation assay and cell proliferation ability was measured by CCK-8 kit. Wound healing experiment was applied to test the cell migration and transwell assay was applied to test the invasion ability. Luciferase assay was employed to demonstrate EWSAT1 and miR-330-5p interaction. In cervical cancer, the expression of EWSAT1 was enhanced and contributed to the poor prognosis. Downregulated EWSAT1 expression inhibited Hela cell migration, proliferation, and invasion. EWSAT1 targeted to miR-330-5p and upregulated cytoplasmic polyadenylation element-binding protein 4 (CPEB4) expression by sponging miR-330-5p. Our study revealed that EWSAT1 enhances CPEB4 expression through sponging miR-330-5p, thereby promoting cervical cancer development, which might provide potential therapeutic targets for clinically cervical cancer patients.

Adversarial Learning for Joint Optimization of Depth and Ego-Motion.

In recent years, supervised deep learning methods have shown a great promise in dense depth estimation. However, massive high-quality training data are expensive and impractical to acquire. Alternatively, self-supervised learning-based depth estimators can learn the latent transformation from monocular or binocular video sequences by minimizing the photometric warp error between consecutive frames, but they suffer from the scale ambiguity problem or have difficulty in estimating precise pose changes between frames. In this paper, we propose a joint self-supervised deep learning pipeline for depth and ego-motion estimation by employing the advantages of adversarial learning and joint optimization with spatial-temporal geometrical constraints. The stereo reconstruction error provides the spatial geometric constraint to estimate the absolute scale depth. Meanwhile, the depth map with an absolute scale and a pre-trained pose network serves as a good starting point for direct visual odometry (DVO). DVO optimization based on spatial geometric constraints can result in a fine-grained ego-motion estimation with the additional backpropagation signals provided to the depth estimation network. Finally, the spatial and temporal domain-based reconstructed views are concatenated, and the iterative coupling optimization process is implemented in combination with the adversarial learning for accurate depth and precise ego-motion estimation. The experimental results show superior performance compared with state-of-the-art methods for monocular depth and ego-motion estimation on the KITTI dataset and a great generalization ability of the proposed approach.

Real-Time Vehicle Make and Model Recognition with the Residual SqueezeNet Architecture.

Make and model recognition (MMR) of vehicles plays an important role in automatic vision-based systems. This paper proposes a novel deep learning approach for MMR using the SqueezeNet architecture. The frontal views of vehicle images are first extracted and fed into a deep network for training and testing. The SqueezeNet architecture with bypass connections between the Fire modules, a variant of the vanilla SqueezeNet, is employed for this study, which makes our MMR system more efficient. The experimental results on our collected large-scale vehicle datasets indicate that the proposed model achieves 96.3% recognition rate at the rank-1 level with an economical time slice of 108.8 ms. For inference tasks, the deployed deep model requires less than 5 MB of space and thus has a great viability in real-time applications.

Local Tiled Deep Networks for Recognition of Vehicle Make and Model.

Vehicle analysis involves license-plate recognition (LPR), vehicle-type classification (VTC), and vehicle make and model recognition (MMR). Among these tasks, MMR plays an important complementary role in respect to LPR. In this paper, we propose a novel framework for MMR using local tiled deep networks. The frontal views of vehicle images are first extracted and fed into the local tiled deep networks for training and testing. A local tiled convolutional neural network (LTCNN) is proposed to alter the weight sharing scheme of CNN with local tiled structure. The LTCNN unties the weights of adjacent units and then ties the units k steps from each other within a local map. This architecture provides the translational, rotational, and scale invariance as well as locality. In addition, to further deal with the colour and illumination variation, we applied the histogram oriented gradient (HOG) to the frontal view of images prior to the LTCNN. The experimental results show that our LTCNN framework achieved a 98% accuracy rate in terms of vehicle MMR.