Characteristic Appearance Research Articles

ReGR: Relation-aware graph reasoning framework for video question answering

As one of the challenging cross-modal tasks, video question answering (VideoQA) aims to fully understand video content and answer relevant questions. The mainstream approach in current work involves extracting appearance and motion features to characterize videos separately, ignoring the interactions between them and with the question. Furthermore, some crucial semantic interaction details between visual objects are overlooked. In this paper, we propose a novel Relation-aware Graph Reasoning (ReGR) framework for video question answering, which first combines appearance–motion and location–semantic multiple interaction relations between visual objects. For the interaction between appearance and motion, we design the Appearance–Motion Block, which is question-guided to capture the interdependence between appearance and motion. For the interaction between location and semantics, we design the Location–Semantic Block, which utilizes the constructed Multi-Relation Graph Attention Network to capture the geometric position and semantic interaction between objects. Finally, the question-driven Multi-Visual Fusion captures more accurate multimodal representations. Extensive experiments on three benchmark datasets, TGIF-QA, MSVD-QA, and MSRVTT-QA, demonstrate the superiority of our proposed ReGR compared to the state-of-the-art methods.

MERASTC: Micro-Expression Recognition Using Effective Feature Encodings and 2D Convolutional Neural Network

Facial micro-expression (ME) can disclose genuine and concealed human feelings. It makes MEs extensively useful in real-world applications pertaining to affective computing and psychology. Unfortunately, they are induced by subtle facial movements for a short duration of time, which makes the ME recognition, a highly challenging problem even for human beings. In automatic ME recognition, the well-known features encode either incomplete or redundant information, and there is a lack of sufficient training data. The proposed method, Micro-Expression Recognition by Analysing Spatial and Temporal Characteristics, MERASTC mitigates these issues for improving the ME recognition. It compactly encodes the subtle deformations using action units (AUs), landmarks, gaze, and appearance features of all the video frames while preserving most of the relevant ME information. Furthermore, it improves the efficacy by introducing a novel neutral face normalization for ME and initiating the utilization of gaze features in deep learning based ME recognition. The features are provided to the 2D convolutional neural network that jointly analyses the spatial and temporal behavior for correct ME classification. Experimental results on the publicly available datasets indicate that the proposed method exhibits better performance than the well-known methods.

A CNN-Based Facial Expression Recognition System

Facial expression recognition has long attracted researchers from a variety of domains, such as human-computer interaction, emotion analysis, intelligent medical care, and so on. Various human-designed features have been used in FER to extract image appearance features, but human-designed methods are difficult to be used to recognize new face images, which poses challenges for FER in uncontrolled environments. Recently, CNNs and FNNs have been used for facial expression recognition. The new CNN-based recognition approach yields excellent FER results. The numerous convolution and pooling layers that CNN possesses allow it to extract higher and multi-level features of the entire face or local regions. Additionally, CNN has excellent classification performance when it comes to the feature extraction of facial expression pictures. However, the mechanism through which FNN remains largely controversial. In this work, the experiments are carried out on the comparison of three facial expression recognition algorithms: K-Nearest Neighbor (KNN), Support Vector Machine (SVM), and Convolutional Neural Network (CNN), and the experimental results show that, using CNN can get more stable recommendation results of better quality.

Efficacy and Tolerability of Cosmetic Serums Enriched with Five Forms of Hyaluronic Acid as Part of Biweekly Diamond Tip Microdermabrasion Treatments for Facial Skin Dryness and Age-Associated Features.

There is growing interest in combining topical treatments with aesthetic procedures to combat signs of aging skin. This study aimed to assess the efficacy and tolerability of a novel cosmetic serum containing 5 different forms of HA (HA5 DG) when used via a proprietary diamond-tip microdermabrasion procedure (DG) to treat skin dryness, fine lines/wrinkles, rough texture, and dullness. In this open-label, single-center study, participants received HA5 DG as part of a biweekly DG procedure on the face and neck for 12 weeks. Study participants also applied another take-home HA5 serum to the face twice daily at home, along with a basic skincare regimen. The efficacy of the combined treatment was measured by clinical quantification of multiple skin appearance features, analysis of bioinstrumental measurements, and digital photography. This study enrolled 27 participants, with an average age of 42.7 years and Fitzpatrick skin phototypes I-III (59.3%), IV (18.5%), and V-VI (22.2%), and 23 participants completed the study. The combined treatment had effects in fine lines/wrinkles, skin dryness, smoothness, radiance, firmness, and hydration 15 minutes post-DG. Furthermore, the significant improvements observed in dryness, fine lines/wrinkles, skin smoothness, and radiance were still visible 3 days after and maintained at week 12. Additionally, smoothing of coarse lines/wrinkles, improvement of skin tone evenness, hyperpigmentation, photodamage, and transepidermal water loss were observed at week 12. The treatment had a favorable tolerability profile and was perceived as efficacious and highly satisfactory. This novel combined treatment delivered immediate and prolonged skin hydration and high participant satisfaction, proving it can be an excellent approach for skin rejuvenation.

PSNet: Parallel Symmetric Network for Video Salient Object Detection

For the video salient object detection (VSOD) task, how to excavate the information from the appearance modality and the motion modality has always been a topic of great concern. The two-stream structure, including an RGB appearance stream and an optical flow motion stream, has been widely used as a typical pipeline for VSOD tasks, but the existing methods usually only use motion features to unidirectionally guide appearance features or adaptively but blindly fuse two modality features. However, these methods underperform in diverse scenarios due to the uncomprehensive and unspecific learning schemes. In this paper, following a more secure modeling philosophy, we deeply investigate the importance of appearance modality and motion modality in a more comprehensive way and propose a VSOD network with up and down parallel symmetry, named PSNet. Two parallel branches with different dominant modalities are set to achieve complete video saliency decoding with the cooperation of the Gather Diffusion Reinforcement (GDR) module and Cross-modality Refinement and Complement (CRC) module. Finally, we use the Importance Perception Fusion (IPF) module to fuse the features from two parallel branches according to their different importance in different scenarios. Experiments on four dataset benchmarks demonstrate that our method achieves desirable and competitive performance.

Attention correlated appearance and motion feature followed temporal learning for activity recognition

<span lang="EN-US">Recent advances in deep neural networks have been successfully demonstrated with fairly good accuracy for multi-class activity identification. However, existing methods have limitations in achieving complex spatial-temporal dependencies. In this work, we design two stream fusion attention (2SFA) connected to a temporal bidirectional gated recurrent unit (GRU) one-layer model and classified by prediction voting classifier (PVC) to recognize the action in a video. Particularly in the proposed deep neural network (DNN), we present 2SFA for capturing appearance information from red green blue (RGB) and motion from optical flow, where both streams are correlated by proposed fusion attention (FA) as the input of a temporal network. On the other hand, the temporal network with a bi-directional temporal layer using a GRU single layer is preferred for temporal understanding because it yields practical merits against six topologies of temporal networks in the UCF101 dataset. Meanwhile, the new proposed classifier scheme called PVC employs multiple nearest class mean (NCM) and the SoftMax function to yield multiple features outputted from temporal networks, and then votes their properties for high-performance classifications. The experiments achieve the best average accuracy of 70.8% in HMDB51 and 91.9%, the second best in UCF101 in terms of 2DConvNet for action recognition.</span>

Friction Stir-Spot Welding of AA5052-H32 Alloy Sheets: Effects of Dwell Time on Mechanical Properties and Microstructural Evolution

Friction stir-spot welding (FSSW) as a solid-state joining process for local welding offers a number of benefits for applications in the automotive, aerospace, and marine industries. In these industries, and from an economic point of view, producing spot welds at a low rotating speed and in a short time is critical for saving energy and enhancing productivity. This investigation helped fill a knowledge gap in the literature about FSSW of 4 mm similar lap joints of AA5052-H32 sheet materials, in which welding takes place over a short time period with a slow tool rotation speed. Consequently, the purpose of this work was to investigate the feasibility of FSSW 2 mm thick AA5052-H32 aluminum alloy sheets to produce 4 mm thick similar spot lap joints at various low dwell times of 1, 2, and 3 s and a constant relatively low tool rotation speed of 500 rpm. The introduced heat input for the friction stir-spot welded (FSSWed) lap joints was calculated based on the applied processing parameters. Joint appearance, cross-section macrostructures, and microstructure features of all the spot welds were evaluated. The mechanical properties (hardness contour maps and maximum tensile shear loads) were also examined. The results show that joining 2 mm sheet thickness AA5052-H32 at a low heat input in defect-free similar lap joints could be successfully achieved. The stir zone (SZ) region became wider as the dwell time increased from 1 to 3 s. The hardness value of the SZ was higher than that attained by the AA5052-H32 base material (BM) for all applied dwell times. Especially at 2 s, the hardness of the SZ was approximately 48% higher than that of the BM. This increase in hardness may be attributed to the high grain refinement of the new dynamically recrystallized grain (4 µm) in the SZ compared to the cold-rolled BM grain size (40 µm). Among the tried FSSW process variables, the dwell time of 2 s at a rotation rate of 500 rpm also produced the maximum tensile shear load of 4330 N. Finally, the locations and features of the fracture surfaces of the FSSWed joints were examined using a scanning electron microscope (SEM) and the obtained results were discussed.

Early Femoral Insufficiency Fractures After Primary Total Knee Arthroplasty.

Periprosthetic femur fracture following total knee arthroplasty (TKA) is a devastating complication. Although trauma-related periprosthetic femur fractures have been well studied, early atraumatic insufficiency periprosthetic fractures (IPFs) are gaining attention. We present the largest IPF series to date to better understand and prevent this complication. A retrospective study of all patients who underwent a revision surgery for periprosthetic fracture within 6 months of primary TKA between 2007 and 2020 was performed. Patient demographics, preoperative radiographs, implant details, and fracture radiographs were reviewed. Alignment measurements and fracture characteristics were assessed. Sixteen patients met criteria (rate 0.05%), and 11 had posterior-stabilized TKAs. The mean age was 79 years, mean body mass index was 31 kg/m2, and 94% (15/16) were female. Seven (47%) patients had a confirmed history of osteoporosis. IPF occurred on average 4 weeks (range, 4 days-13 weeks) after the index TKA. Overall, 12 of 16 (73%) had preoperative valgus deformities, and 11 patients (10 valgus, 1 varus) had preoperative deformities >10 degrees. A characteristic radiographic appearance of femoral condylar impaction and collapse was noted in 12 of 16 cases (75%); 11 of these 12 fractures (92%) involved the unloaded compartment based on preoperative varus/valgus deformity. Patients who developed IPFs were most commonly elderly, obese women with osteoporosis and severe preoperative valgus deformities. The apparent mechanism of failure was overloading of previously unloaded osteopenic femoral condyle. In high-risk patients, the use of a cruciate-retaining femoral component or a femoral stem for a posterior-stabilized femur may be considered to help avoid this catastrophic complication.

Lip synechiae: A sequelae of carbamazepine‐induced erythema multiforme. A case report and review of current perspectives and controversies

AbstractLip synechiae as a sequalae of erythema multiforme (EM) is rare. However, the condition results in significant dysfunctionality and disfigurement in affected individuals. The objective of this clinical case report is to highlight the importance of the preventative strategies in carbamazepine‐induced EM to alleviate complications such as lip synechiae. We reviewed a case report of lip synechiae as a sequelae of carbamazepine‐induced EM as well as the previously reported cases of lip synechiae. A 23‐year‐old female presented with the inability to open her mouth following carbamazepine exposure with concomitant eruption of mucocutaneous lesions on the torso, acral, conjunctiva and haemorrhagic crusting on both lips with fusion thereof. The lip synechiae resulted in dysfunction and disfigurement with reduced verbal articulation and mastication. Based on the history and the characteristic appearance of the mucocutaneous lesions, a diagnosis of EM was made. The suspected inciting drug was withdrawn followed by bilateral chalinoplasty. To prevent further adhesion of the lips, chlorhexidine‐impregnated paraffin gauze dressing was applied. The patient attained good functional and aesthetic outcomes.

A Review of Techniques on Gait-Based Person Re-Identification

Survey/review study A Review of Techniques on Gait-Based Person Re-Identification Babak Rahi 1,*, Maozhen Li 1, and Man Qi 2 1 Department of Electronics and Computer Engineering, Brunel University London, Uxbridge, Middlesex, UB8 3PH, United Kingdom 2 The School of Engineering, University of Warwick, Coventry CV4 7AL, United Kingdom * Correspondence: babak.h.rahi@hotmail.com Received: 16 October 2022 Accepted: 14 December 2022 Published: 27 March 2023 Abstract: Person re-identification at a distance across multiple non-overlapping cameras has been an active research area for years. In the past ten years, short-term Person re-identification techniques have made great strides in accuracy using only appearance features in limited environments. However, massive intra-class variations and inter-class confusion limit their ability to be used in practical applications. Moreover, appearance consistency can only be assumed in a short time span from one camera to the other. Since the holistic appearance will change drastically over days and weeks, the technique, as mentioned above, will be ineffective. Practical applications usually require a long-term solution in which the subject's appearance and clothing might have changed after the elapse of a significant period. Facing these problems, soft biometric features such as Gait has stirred much interest in the past years. Nevertheless, even Gait can vary with illness, ageing and emotional states, walking surfaces, shoe types, clothes types, carried objects (by the subject) and even environment clutters. Therefore, Gait is considered as a temporal cue that could provide biometric motion information. On the other hand, the shape of the human body could be viewed as a spatial signal which can produce valuable information. So extracting discriminative features from both spatial and temporal domains would benefit this research. This article examines the main approaches used in gait analysis for re-identification over the past decade. We identify several relevant dimensions of the problem and provide a taxonomic analysis of current research. We conclude by reviewing the performance levels achievable with current technology and providing a perspective on the most challenging and promising research directions.

Based on 1H NMR and LC-MS metabolomics reveals biomarkers with neuroprotective effects in multi-parts ginseng powder

In the absence of appearance features, it is becoming increasingly difficult to accurately distinguish botanical herbal powders from different families or parts of the plant. In this study, we explored the biomarkers with neuroprotective effects in multi-parts ginseng based on metabolite content changes and biological functions. First, using 1H NMR and LC-MS metabolomics techniques and multivariate statistical analysis techniques to analyze the differential metabolites (DMs) in ginseng main root (TR), lateral root (LR) and Rhizome (RZ). Also, the analysis of 14 differential metabolites (DMs) using network pharmacology yielded 23 core targets associated with neuroprotective diseases, the KEGG pathway was enriched for 117 signaling pathways and 184 enrichment processes were shown in the GO analysis. Combined with the correlation analysis of primary differential metabolites (PDMs) and secondary differential metabolites (SDMs) content and network activity, we finally identified three biomarkers in different parts of ginseng, namely sucrose, ginsenoside Rb1, and ginsenoside Rb3. Finally, the three biomarkers were validated by molecular docking and SH-SY5Y survival, and the results showed that sucrose, ginsenoside Rb1 and ginsenoside Rb3 exhibited favorable activity results in neuroprotective functions. The results of this study reveal the overall differences in the DMs of the three parts of ginseng, and three biomarkers can help identify different parts of ginseng powder, to avoid illegal substitution when ginseng is used in medicine. At the same time, it also provides a reference method for screening biomarkers that can distinguish different parts of traditional Chinese medicine.

Study on the phase structure of the interface zone of Cu–Al composite plate in cast-rolling state and different heat treatment temperatures based on EBSD

The phase structure in the interface zone of Cu–Al composite plate was studied by EBSD after casting and rolling and different heat treatment temperatures. The grain sizes of the matrix and intermetallic compounds in the interface zone were analyzed statistically. The spatial distribution characteristics of the transverse, rolled, normal and various phases were characterized by inverse pole diagram. The results show that with the increase of heat treatment temperature, the phase of the interfacial area of the composite plate is rearranged, the residual stress can be released, and the zero resolution decreases. The interface zone of cast-rolled Cu–Al composite plate is composed of a layer of intermetallic compound, which is Al2Cu; Two layers of intermetallic compounds, both Al, were formed in the interfacial area after heat treatment at 300 °C and 400 °C Al2Cu and Al4Cu9;After heat treatment at 450 °C, the interface region is composed of three layers of intermetallic compounds, which is Al2Cu, AlCu, Al4Cu9;The grain size of copper matrix is obviously larger than that of aluminum matrix. Matrix aluminum, matrix copper, Al2Cu.The average grains of Cu are mostly large Angle grain boundaries, and the orientation difference of adjacent grains in aluminum layer is much smaller than that in copper layer. After heat treatment at 300 °C for 1 h, the recrystallization degree of the interface zone of Cu–Al composite plate is low. After heat treatment at 400 °C for 1 h, the recrystallization in the interfacial zone of Cu–Al composite plate is not complete; After annealing at 450 °C for 1 h, the subgrain boundaries in the aluminum layer grow up, and LAGB is basically converted to HAGB. The distribution characteristics of the characteristic appearance direction in the crystal space of each phase were characterized under the conditions of casting and rolling and different heat treatment temperatures. At the same time, the research results provide theoretical basis and scientific title for optimizing the preparation process and high-end application of Cu–Al composite plate.

Knowledge and Attitude of Students at Ardabil University of Medical Sciences Regarding Hygiene and Food Safety During the COVID-19 Pandemic

Background: Coronaviridae is a peculiar viral family with a vast ribonucleic acid (RNA) genome and characteristic appearance, endowed with an astonishing tendency to transfer from animals to humans. Various factors can affect preventing this emerging disease and controlling its severity. Objectives: This study aimed to evaluate the knowledge and attitude of students of Ardabil University of Medical Science (ARUMS) bout hygiene and food safety and their association with COVID-19 disease. Methods: This cross-sectional and descriptive study used an online questionnaire by random sampling, and the data were analyzed with SPSS software version 22. In this study, the statistical population was the students of ARUMS, and 100 (59 females and 41 males) students were selected from all the university students. A non-parametric test (Kruskal Wallis, Mann-Whitney), chi-square and Student’s t-test were used for statistical analysis. Results: There was a significant correlation between gender (male and female) and the level of understanding in three categories (weak, moderate, and good) (P = 0.048). However, there was no significant relationship with the variable of attitude. Determining the correlation between the variables of students’ knowledge and perspective with the Pearson correlation test revealed that the intersection between hygiene and food safety variables with the variables of awareness, attitude, and COVID-19 is significant. Conclusions: Having appropriate behaviors and knowledge about eating habits is essential for medical students who are constantly faced with high-risk environments regarding pathogens and need to strengthen their immune system and keep it at the desired level.

Multi-Target Tracking Based on a Combined Attention Mechanism and Occlusion Sensing in a Behavior-Analysis System.

Multi-object tracking (MOT) is a topic of great interest in the field of computer vision, which is essential in smart behavior-analysis systems for healthcare, such as human-flow monitoring, crime analysis, and behavior warnings. Most MOT methods achieve stability by combining object-detection and re-identification networks. However, MOT requires high efficiency and accuracy in complex environments with occlusions and interference. This often increases the algorithm's complexity, affects the speed of tracking calculations, and reduces real-time performance. In this paper, we present an improved MOT method combining an attention mechanism and occlusion sensing as a solution. A convolutional block attention module (CBAM) calculates the weights of space and channel attention from the feature map. The attention weights are used to fuse the feature maps to extract adaptively robust object representations. An occlusion-sensing module detects an object's occlusion, and the appearance characteristics of an occluded object are not updated. This can enhance the model's ability to extract object features and improve appearance feature pollution caused by the short-term occlusion of an object. Experiments on public datasets demonstrate the competitive performance of the proposed method compared with the state-of-the-art MOT methods. The experimental results show that our method has powerful data association capability, e.g., 73.2% MOTA and 73.9% IDF1 on the MOT17 dataset.

Behavior-Based Video Summarization System for Dog Health and Welfare Monitoring

The popularity of dogs has been increasing owing to factors such as the physical and mental health benefits associated with raising them. While owners care about their dogs' health and welfare, it is difficult for them to assess these, and frequent veterinary checkups represent a growing financial burden. In this study, we propose a behavior-based video summarization and visualization system for monitoring a dog's behavioral patterns to help assess its health and welfare. The system proceeds in four modules: (1) a video data collection and preprocessing module; (2) an object detection-based module for retrieving image sequences where the dog is alone and cropping them to reduce background noise; (3) a dog behavior recognition module using two-stream EfficientNetV2 to extract appearance and motion features from the cropped images and their respective optical flow, followed by a long short-term memory (LSTM) model to recognize the dog's behaviors; and (4) a summarization and visualization module to provide effective visual summaries of the dog's location and behavior information to help assess and understand its health and welfare. The experimental results show that the system achieved an average F1 score of 0.955 for behavior recognition, with an execution time allowing real-time processing, while the summarization and visualization results demonstrate how the system can help owners assess and understand their dog's health and welfare.

Abnormal Behavior Monitoring Method of Larimichthys crocea in Recirculating Aquaculture System Based on Computer Vision.

It is crucial to monitor the status of aquaculture objects in recirculating aquaculture systems (RASs). Due to their high density and a high degree of intensification, aquaculture objects in such systems need to be monitored for a long time period to prevent losses caused by various factors. Object detection algorithms are gradually being used in the aquaculture industry, but it is difficult to achieve good results for scenes with high density and complex environments. This paper proposes a monitoring method for Larimichthys crocea in a RAS, which includes the detection and tracking of abnormal behavior. The improved YOLOX-S is used to detect Larimichthys crocea with abnormal behavior in real time. Aiming to solve the problems of stacking, deformation, occlusion, and too-small objects in a fishpond, the object detection algorithm used is improved by modifying the CSP module, adding coordinate attention, and modifying the part of the structure of the neck. After improvement, the AP50 reaches 98.4% and AP50:95 is also 16.2% higher than the original algorithm. In terms of tracking, due to the similarity in the fish's appearance, Bytetrack is used to track the detected objects, avoiding the ID switching caused by re-identification using appearance features. In the actual RAS environment, both MOTA and IDF1 can reach more than 95% under the premise of fully meeting real-time tracking, and the ID of the tracked Larimichthys crocea with abnormal behavior can be maintained stably. Our work can identify and track the abnormal behavior of fish efficiently, and this will provide data support for subsequent automatic treatment, thus avoiding loss expansion and improving the production efficiency of RASs.

A rare case of Coffin–Lowry syndrome accompanied by a copper-beaten skull appearance

Coffin-Lowry syndrome (CLS), usually a result of loss-of-function mutations in the RPS6KA3 gene located at Xp22.2, is a very rare genetic condition. It is associated with different phenotypes, including dysmorphic facial features, neuro-developmental impairment, short stature, and skeletal deformities. Here we describe a four-year-old boy who had distinct clinical features of CLS: moderate mental and motor retardation, brachycephaly, microcephaly, hypertelorism, low-set and prominent ears, down- slanted palpebral fissures, a depressed nasal bridge with thick nasal alae and septum, a high-arched palate, widely spaced teeth, and retrognathia. Genetic analysis revealed a pathogenic variant in the RPS6KA3 gene, NM_004586.3: exon 22: c.2186G>A: (p.Arg729Gln). His healthy mother was a heterozygous carrier of the variant. In addition to the clinical findings, the patient’s X-ray revealed a very rare feature of a copper-beaten skull appearance. The link between CLS and a copper-beaten skull has never been reported before. This case demonstrates the variability in presentation of CLS and signifies the association of a rare radiographic finding of copper-beaten skull with CLS. Based on this case, we recommend screening skull radiographs at the clinic for copper-beaten skulls in patients with CLS.

Chinese expert consensus on the treatment of MMD

Moyamoya disease (MMD), also known as spontaneous occlusion of the circle of Willis, is defined by progressive stenosis or occlusion of the internal carotid arteries, and it can progress to the anterior, middle, and posterior cerebral arteries. As these arteries are gradually stenosed, a collateral network of capillaries develops at the base of the brain, producing the characteristic reticulate appearance (“puff of smoke”) on angiography. Therefore, it was named by Suzuki and Takaku in 1969 after the Japanese term “moyamoya” (Suzuki and Takaku, Arch Neurol 20:288–299, 1969). MMD is most common in East Asian countries such as Japan and Korea, and it shows a slight female predominance. MMD is mainly characterized by ischemia and hemorrhage. Hemorrhagic MMD is very rare in children, and most cases occur in adults due to the rupture of the compensatory blood vessels, which often leads to hemorrhagic symptoms (Scott and Smith, N Engl J Med 360:1226–1237, 2009). In recent years, the diagnosis rate has increased with the popularization of imaging techniques. However, the pathogenesis of MMD is still not completely understood, and there is currently no evidence to suggest that drug treatment can delay or even reverse the progression of MMD. The current drug treatment for in MMD only targets its clinical symptoms, including ischemia and hemorrhage. The main choice of treatment for MMD is surgical revascularization. As an increasing number of hospitals have developed surgical treatment for MMD, our compiling group has jointly discussed the formulation of a consensus among Chinese experts on the treatment of MMD.

A Computational Meta-Learning Inspired Model for Sketch-based Video Retrieval

Objectives: To design and develop an efficient computing framework for sketch-based video retrieval using fine-grained intrinsic computational approach. Methods: The primary method of sketch-based video retrieval adopts multi-stream multi-modality of joint embedding method for improved P-SBVR from improved fine-grained KTH and TSF related dataset. It considers the potential aspects of the computation of significant visual intrinsic appearance details for sketch objects. The extracted appearance and motion-based features are used to train three different CNN baselines under strong and weak supervision. The system also implements a meta-learning model for different supervised settings to attain better performance of sketch-based video retrieval along with a relational module to overcome the problem of overfitting. Findings: The study derives specific sketch sequences from its formulated dataset to compute instance-level query processing for video retrieval. Further, it also addresses the limitations arising in the context of coarse-grained video retrieval models and sketch-based still image retrieval. The aggregated dataset for rich annotation assisted in the experimental simulation. The experimental evaluation with respect to the performance metric evaluates the 3D CNN baselines under strong supervision and weak-supervision where CNN BL-Type-2 attains maximum video retrieval accuracy of 99.96% for triplet grading feature under relational schema. CNN BL-Type-1 attains maximum retrieval accuracy of 97.40% considering the triplet grading features from the improved SBVR. The evaluation metric for the instance level retrieval process also considers true matching of sketches with the videos, it clearly shows that the appropriate appearance and motion based feature selection has enhanced the video retrieval accuracy up to 96.90% with 99.28% accuracy in action identification considering motion stream, 98.17% for appearance module and 98.45% for fusion module. Another important aspect of the proposed research context is that it addresses the problem of cross-modality while executing the simultaneous matching paradigm for visual appearances of the object with its movement appearing on particular video scenes. The experimental outcome showsits comparable effectiveness relative to the existing system of CNN. Novelty: Unlike the conventional system of sketch analysis, which is more focused on static objects or scenes, the presented approach can efficiently compute the important visual intrinsic appearance details of the object of interest from the sketch and then activate the operations for video retrieval. The proposed CNN based learning model with improved P-SBVR dataset attains better computing time for retrieval with are approximately (200, 210 and 214) milliseconds for CNN BL-Type-1, CNN BL-Type-2, CNN BL-Type-3 and comparable with the existing deep learning based SBVR models. Keywords: Sketch Based Video Retrieval; Intrinsic Appearance Details; Meta Learning; Sketch Dataset; Cross Modality Problem

Exudate identification in retinal fundus images using precise textural verifications

One of the most salient diseases of retina is Diabetic Retinopathy (DR) which may lead to irreparable damages to eye vision in the advanced phases. A large number of the people infected with diabetes experience DR. The early identification of DR signs facilitates the treatment process and prevents from blindness. Hard Exudates (HE) are bright lesions appeared in retinal fundus images of DR patients. Thus, the detection of HEs is an important task preventing the progress of DR. However, the detection of HEs is a challenging process due to their different appearance features. In this paper, an automatic method for the identification of HEs with various sizes and shapes is proposed. The method works based on a pixel-wise approach. It considers several semi-circular regions around each pixel. For each semi-circular region, the intensity changes around several directions and non-necessarily equal radiuses are computed. All pixels for which several semi-circular regions include considerable intensity changes are considered as the pixels located in HEs. In order to reduce false positives, an optic disc localization method is proposed in the post-processing phase. The performance of the proposed method has been evaluated on DIARETDB0 and DIARETDB1 datasets. The experimental results confirm the improved performance of the suggested method in term of accuracy.