Training Photos Research Articles

Development of Attentive-Based Trans-UNet and Cycle Generative Adversarial Networks for Underwater Image Enhancement

Researchers across the globe have been investigating underwater photographs and a way to capture pictures with outstanding clarity for the past couple of decades. Also, improving the obtained photographs is an exhausting task. Usually, applied underwater image-capturing devices are unable to acquire high-resolution photos underwater, and their maintenance is extremely costly. The underwater images include multiple flaws because of biological processes like attenuation and scattering. These photographs experience color distortion, blurriness, and low contrast effects. Technologies that utilize deep learning have become more popular among several research studies and have gradually grown in impact on society. Several techniques demand sets of training photos, but gathering such expected sets can be challenging because of the complex nature of the underwater environment. Generating and restoring an image from water is a difficult task that has gained prominence in recent days. By lowering graininess, adjusting, and refining the photos using deep learning models, the major goal is to enhance underwater images. To accomplish this objective, an intelligent attention-based deep learning model is proposed. In the first stage, the unrefined images are gathered from typical data sources. Further, the collected underwater images are fed into the model of Attentive-based Trans-UNet-CycleGAN (ATUNet-CGAN), where the Transformer-based UNet model is integrated with the Cycle Generative Adversarial Networks (GANs). Also, the attention mechanism process is involved in Trans-UNet-CycleGAN for improving the superiority of submarine images. Finally, the performance of the model is validated using different metrics and correlated among baseline approaches. Therefore, the proposed methodology outperforms the exploitation of better enhancement of image quality.

Federated Learning with Homomorphic Encryption for Ensuring Privacy in Medical Data

Federated Learning (FL) is a machine learning methodology that allows remote devices to collectively train a learning system without sharing their data. FL-based methods provide enhanced secure privacy by transmitting only localized model variables, learned with local information, from dispersed devices to a centralized controller. There is a potential for a centralized server or malicious individuals to deduce or get sensitive private data by analyzing the structure and variables of regional learning networks. This study incorporates the FL process into the deep learning process of medical prototypes in an Internet of Things (IoT)-based medical facility. A Secured Medical Homomorphic Encryption Algorithm (SMHEA) is proposed in this research to ensure medical data privacy. Cryptographic primitives, such as masking and homomorphic cryptography, are used to enhance the security of local modeling. This prevents adversaries from deducing confidential health information via assaults like model restoration or modeling inversion. The primary determinant for assessing the regional modeling's contributions to the universal model during each training stage is the quality of the databases possessed by various individuals rather than the typically used metric of database dimension in deep learning. A dropout-tolerant approach is suggested, where the FL procedure would continue as long as the total amount of online customers remains over a certain level. By doing a security evaluation, it is evident that the suggested approach effectively ensures data privacy. Theoretical analysis is conducted on accuracy, computation time, and communication error. An example of clinical applications is the categorization of skin lesions using training photos from the HAM10000 medical database. The experimental findings demonstrate that the suggested system had favorable performance and privacy preservation outcomes compared to current methods.

Sign Language Recognition by Image Processing

A prototype sign language interpreter that can converse in American Sign Language (ASL) is presented in this study. The visual stimulation of sign language aids in the development of speech and language. It promotes social contact, fosters the development of cognitive structures, and lessens harmful social behaviors. Early exposure to sign language supports language development in academic, social, and emotional domains. Building a user-hand gesture-only human-computer interface (HCI) is the foundation of the dependable paradigm. Using Python and OpenCV, hand gesture recognition is built using the theories of hand segmentation and the hand detection system. The training photos are labeled with different names based on what is in the database. A binary feature vector is created from each image. Convolutional neural networks (CNNs) and cameras are used to capture images from hand sign videos in real time, which are then used to carry out the feature extraction process. The algorithm generates output based on the projected maximum similarity between these features and the features of database photographs. With the aid of an application, this model provides adequate accuracy even in the absence of a constant or monochromatic background. The person's hand movements are captured on video with the camera; the video is subsequently processed and translated into text. The user hears the speech that has been transformed from the output. The suggested system recognizes letters, numbers, and certain common words in sign language. Raw photos and videos are transformed into corresponding text that can be read and comprehended by using image processing algorithms and neural networks to map the gesture to the right text in the training data. When paired with a sizable source database, this protocol will surely be highly beneficial for closing the communication gap between people who can talk and hear and those who cannot.

Assessment of Narrow-Band Imaging Algorithm for Video Capsule Endoscopy Based on Decorrelated Color Space for Esophageal Cancer: Part II, Detection and Classification of Esophageal Cancer

Esophageal carcinoma (EC) is a prominent contributor to cancer-related mortality since it lacks discernible features in its first phases. Multiple studies have shown that narrow-band imaging (NBI) has superior accuracy, sensitivity, and specificity in detecting EC compared to white light imaging (WLI). Thus, this study innovatively employs a color space linked to décor to transform WLIs into NBIs, offering a novel approach to enhance the detection capabilities of EC in its early stages. In this study a total of 3415 WLI along with the corresponding 3415 simulated NBI images were used for analysis combined with the YOLOv5 algorithm to train the WLI images and the NBI images individually showcasing the adaptability of advanced object detection techniques in the context of medical image analysis. The evaluation of the model’s performance was based on the produced confusion matrix and five key metrics: precision, recall, specificity, accuracy, and F1-score of the trained model. The model underwent training to accurately identify three specific manifestations of EC, namely dysplasia, squamous cell carcinoma (SCC), and polyps demonstrates a nuanced and targeted analysis, addressing diverse aspects of EC pathology for a more comprehensive understanding. The NBI model effectively enhanced both its recall and accuracy rates in detecting dysplasia cancer, a pre-cancerous stage that might improve the overall five-year survival rate. Conversely, the SCC category decreased its accuracy and recall rate, although the NBI and WLI models performed similarly in recognizing the polyp. The NBI model demonstrated an accuracy of 0.60, 0.81, and 0.66 in the dysplasia, SCC, and polyp categories, respectively. Additionally, it attained a recall rate of 0.40, 0.73, and 0.76 in the same categories. The WLI model demonstrated an accuracy of 0.56, 0.99, and 0.65 in the dysplasia, SCC, and polyp categories, respectively. Additionally, it obtained a recall rate of 0.39, 0.86, and 0.78 in the same categories, respectively. The limited number of training photos is the reason for the suboptimal performance of the NBI model which can be improved by increasing the dataset.

A Literature Survey on Completely Blind Image Quality Evaluator Feature-Enriched

Existing blind image quality assessment (BIQA) technique in which are principally opinion-aware. They learn deterioration models from training photos with associated human subjective scores to predict the sensory activity quality of take a look at pictures. Such opinion-aware methods, however, need a bulky amount of training example with associated human prejudiced achieve moreover of a variety of deformation category. The BIQA representation find out through opinion-aware strategies typically have weak generalization capability, herewith limiting their usability in observe. By comparison, opinion-unaware strategies don't require human prejudiced achieve for training, and then have superior possible designed for good quality simplification ability. Unluckily, to the present point no opinion-unaware BIQA technique has shown systematically better quality prediction accuracy than the opinion-aware strategies. Here, we tend to aim to develop an opinionunaware BIQA technique that may compete with, and maybe exceed, the present opinion-aware methods. By integration the options of ordinary picture information consequent as of multiple cues, we have an affinity to discover a multivariate Gaussian structure of representation patches from a set of perfect natural pictures. using the learned multivariate Gaussian form, a Bhattacharyya-like distance is employed to measure the standard of every image patch, and then a generally value score is find by average pooling. The projected BIQA technique doesn't want any distorted sample images or subjective class achieve for training, yet extensive experiment show its better quality-prediction presentation to the situation of the art opinion-aware BIQA strategies.

Effective texture features in mammogram images via multi-roi segmentation

Digital mammography increasingly necessitates image segmentation for the purpose of dividing mammograms into individual slices. For the purpose of removing suspicious masses or tumours from mammograms, this process is carried out using a region of interest (ROI). More training photos are needed for mammography image classification, and these circumstances, ROI requires more processing time. The temporal complexity difficulties with the suggested multi-ROI method are the subject of this article. To show how effective the suggested multi-ROI is compared to the current segmentation approach, experiments are conducted on benchmarked datasets.

A Literature Survey on Completely Blind Image Quality Evaluator Feature-Enriched

Existing blind image quality assessment (BIQA) technique in which are principally opinion-aware. They learn deterioration models from training photos with associated human subjective scores to predict the sensory activity quality of take a look at pictures. Such opinion-aware methods, however, need a bulky amount of training example with associated human prejudiced achieve moreover of a variety of deformation category. The BIQA representation find out through opinion-aware strategies typically have weak generalization capability, herewith limiting their usability in observe. By comparison, opinion-unaware strategies don't require human prejudiced achieve for training, and then have superior possible designed for good quality simplification ability. Unluckily, to the present point no opinion-unaware BIQA technique has shown systematically better quality prediction accuracy than the opinion-aware strategies. Here, we tend to aim to develop an opinionunaware BIQA technique that may compete with, and maybe exceed, the present opinion-aware methods. By integration the options of ordinary picture information consequent as of multiple cues, we have an affinity to discover a multivariate Gaussian structure of representation patches from a set of perfect natural pictures. using the learned multivariate Gaussian form, a Bhattacharyya-like distance is employed to measure the standard of every image patch, and then a generally value score is find by average pooling. The projected BIQA technique doesn't want any distorted sample images or subjective class achieve for training, yet extensive experiment show its better quality-prediction presentation to the situation of the art opinion-aware BIQA strategies.

An artificial intelligence model to identify snakes from across the world: Opportunities and challenges for global health and herpetology.

BackgroundSnakebite envenoming is a neglected tropical disease that kills an estimated 81,000 to 138,000 people and disables another 400,000 globally every year. The World Health Organization aims to halve this burden by 2030. To achieve this ambitious goal, we need to close the data gap in snake ecology and snakebite epidemiology and give healthcare providers up-to-date knowledge and access to better diagnostic tools. An essential first step is to improve the capacity to identify biting snakes taxonomically. The existence of AI-based identification tools for other animals offers an innovative opportunity to apply machine learning to snake identification and snakebite envenoming, a life-threatening situation.MethodologyWe developed an AI model based on Vision Transformer, a recent neural network architecture, and a comprehensive snake photo dataset of 386,006 training photos covering 198 venomous and 574 non-venomous snake species from 188 countries. We gathered photos from online biodiversity platforms (iNaturalist and HerpMapper) and a photo-sharing site (Flickr).Principal findingsThe model macro-averaged F1 score, which reflects the species-wise performance as averaging performance for each species, is 92.2%. The accuracy on a species and genus level is 96.0% and 99.0%, respectively. The average accuracy per country is 94.2%. The model accurately classifies selected venomous and non-venomous lookalike species from Southeast Asia and sub-Saharan Africa.ConclusionsTo our knowledge, this model’s taxonomic and geographic coverage and performance are unprecedented. This model could provide high-speed and low-cost snake identification to support snakebite victims and healthcare providers in low-resource settings, as well as zoologists, conservationists, and nature lovers from across the world.

Union is Strength: Improving face sketch synthesis by fusing outcomes of Fully-Convolutional-Networks and Random Sampling Locality Constraint

Automated formulation of sketches from face photos has seen successive growth since the work of Wang and Tang in recent years. Each new methodology is, however, able to partially achieve its objective of sketch synthesis while using pairs of photos and viewed sketches as a training medium. The viewed sketches are also used as a testing medium to determine the success of those methodologies. Resulting sketches do not fully capture all features of the training photos and viewed sketches. Their similarity value to respective sketch is also around 30 – 50%. One technique may produce sketches with sharp edges, but they do not bear completeness of facial features. Another technique produces sketches with the completeness of facial elements, but they are not well-focused. Second limitation of existing techniques is attributable to face-recognition procedure which is used as a validation step for these methodologies. Face-recognition process with help of synthesized sketches delivers reliable results over datasets with a limited diversity of age, ethnicity, and light intensities. We propose a novel and cost-effective approach to fuse resulting sketches of two test techniques. The two techniques are merged to yield a better sketch containing well-defined features, sharp contours, and less noise. Secondly, fusion suppresses limitations of the component methodologies reaching the resulting sketch. To test this idea of combining sketch-synthesis methods, we experiment with the most basic techniques of image fusion including simple (arithmetic), PCA, and Wavelet based fusions. The proposed setup considered FCN (complete features but less sharpness) and Fast-RSLCR (sharp edges but missing contours) as candidate techniques. It is tested on two datasets namely CUFS and CUFSF. Second dataset incorporates variations of age, ethnicity, light intensities, and slightly deformed features between photos and viewed sketches. Our results indicate achievement of 60.29% SSIM score (enhancement by 3.84%) and 79.03% face-recognition score (enhancement by 5.62%) as compared to Fast-RSLCR.

CORONAVIRUS Diagnosis Based on Chest X-Ray Images and Pre-trained DenseNet-121

A serious global problem called COVID-19 has killed a great number of people and rendered many projects useless. The obtained individual's identification at the appropriate time is one of the crucial methods to reduce losses. By detecting and recognizing contaminated individuals in the early stages, artificial intelligence can help many associations in these situations. In this study, we offer a fully automated method to identify COVID-19 from a patient's chest X-ray images without the need for a clinical expert's assistance. A new dataset was released, which consists of 300 chest X-ray images from 100 healthy individuals, 100 individuals who were infected with Covid 19, and 100 images of viral pneumonitis. 100 more for testing, too. In order to attain an F1 score of 0.98, a Recall of 0.98, and also an Accuracy of 0.98 with this dataset, a classification method deep learning-based learning algorithm DenseNet-121, transfer learning, as well as data augmentation techniques were implemented. Therefore, even though there are not enough training photos, these findings are far better than other state-of-the-art.

Neural Probabilistic Graphical Model for Face Sketch Synthesis.

Neural network learning for face sketch synthesis from photos has attracted substantial attention due to its favorable synthesis performance. However, most existing deep-learning-based face sketch synthesis models stacked only by multiple convolutional layers without structured regression often lose the common facial structures, limiting their flexibility in a wide range of practical applications, including intelligent security and digital entertainment. In this article, we introduce a neural network to a probabilistic graphical model and propose a novel face sketch synthesis framework based on the neural probabilistic graphical model (NPGM) composed of a specific structure and a common structure. In the specific structure, we investigate a neural network for mapping the direct relationship between training photos and sketches, yielding the specific information and characteristic features of a test photo. In the common structure, the fidelity between the sketch pixels generated by the specific structure and their candidates selected from the training data are considered, ensuring the preservation of the common facial structure. Experimental results on the Chinese University of Hong Kong face sketch database demonstrate, both qualitatively and quantitatively, that the proposed NPGM-based face sketch synthesis approach can more effectively capture specific features and recover common structures compared with the state-of-the-art methods. Extensive experiments in practical applications further illustrate that the proposed method achieves superior performance.

A Superpixel-Wise Approach for Face Sketch Synthesis

Face sketch synthesis from an input photo has drawn great attention in law enforcement and digital entertainment applications. Currently, the input photo is simply divided into overlapped rectangular patches and transformed to a sketch through weighted average of corresponding sketch patches. However, the regular patches lead to defects in the structure of synthesized sketch. In addition, existing methods need to cross the whole training dataset in order to collect the corresponding sketch patches, which limits their usability with the big training datasets. In this paper, a superpixel-wise approach based on the Superpixel technique incorporated into the Locality-constraint Linear Coding (LLC), termed as SuperLLC, is proposed to enhance the facial structure of synthesized sketch and simultaneously maintain fixed computational complexity regardless of the training dataset size. First, the input photo is segmented into overlapped superpixels to find their corresponding sketch superpixels from the training dataset. The LLC is then imposed to regularize proper weights to reconstruct the target sketch from these sketch superpixels, with averaging the overlapped areas between adjacent superpixels. But before all these, for each input photo, a sub-training set is selected based on facial landmarks distance between the input photo and the training photos set. This insures a steady synthesis process. Both subjective and objective experiments on public face sketch databases are finally carried out to reveal the superior performance of the proposed SuperLLC method compared to state-of-the-art methods.

Data-driven vs. model-driven: Fast face sketch synthesis

Face sketch synthesis refers to the technique generating a sketch from an input photo. Existing methods are data-driven, which synthesize a sketch by linearly combining K candidate sketch patches which are purposely selected from the training data. However, these methods have large computation cost due to neighbor selection process that perform neighbor searching on a large scale of training image patches. Instead of the aforementioned commonly used data-driven strategy, we propose to learn some models from training photos to training sketches which could speed the synthesis process a lot while preserving comparable or even better synthesis performance. Specially, we learn some ridge regressors from training photo patch intensities to training sketch patch intensities. An initial estimation is obtained from these regressors. Simultaneously, a high-frequency image is hallucinated from some ridge regressors which are learned from the high-frequency information of training photos and sketches. The high-frequency image is superimposed to the initial estimation to compensate the filtered details due to the dense average in the initial estimation process. Extensive experiments on public face sketch database illustrate the effectiveness of proposed model-driven strategy.

Semantic Photo Retargeting Under Noisy Image Labels

With the popularity of mobile devices, photo retargeting has become a useful technique that adapts a high-resolution photo onto a low-resolution screen. Conventional approaches are limited in two aspects. The first factor is the de-emphasized role of semantic content that is many times more important than low-level features in photo aesthetics. Second is the importance of image spatial modeling: toward a semantically reasonable retargeted photo, the spatial distribution of objects within an image should be accurately learned. To solve these two problems, we propose a new semantically aware photo retargeting that shrinks a photo according to region semantics. The key technique is a mechanism transferring semantics of noisy image labels (inaccurate labels predicted by a learner like an SVM) into different image regions. In particular, we first project the local aesthetic features (graphlets in this work) onto a semantic space, wherein image labels are selectively encoded according to their noise level. Then, a category-sharing model is proposed to robustly discover the semantics of each image region. The model is motivated by the observation that the semantic distribution of graphlets from images tagged by a common label remains stable in the presence of noisy labels. Thereafter, a spatial pyramid is constructed to hierarchically encode the spatial layout of graphlet semantics. Based on this, a probabilistic model is proposed to enforce the spatial layout of a retargeted photo to be maximally similar to those from the training photos. Experimental results show that (1) noisy image labels predicted by different learners can improve the retargeting performance, according to both qualitative and quantitative analysis, and (2) the category-sharing model stays stable even when 32.36% of image labels are incorrectly predicted.

Perceptually Guided Photo Retargeting.

We propose perceptually guided photo retargeting, which shrinks a photo by simulating a human's process of sequentially perceiving visually/semantically important regions in a photo. In particular, we first project the local features (graphlets in this paper) onto a semantic space, wherein visual cues such as global spatial layout and rough geometric context are exploited. Thereafter, a sparsity-constrained learning algorithm is derived to select semantically representative graphlets of a photo, and the selecting process can be interpreted by a path which simulates how a human actively perceives semantics in a photo. Furthermore, we learn the prior distribution of such active graphlet paths (AGPs) from training photos that are marked as esthetically pleasing by multiple users. The learned priors enforce the corresponding AGP of a retargeted photo to be maximally similar to those from the training photos. On top of the retargeting model, we further design an online learning scheme to incrementally update the model with new photos that are esthetically pleasing. The online update module makes the algorithm less dependent on the number and contents of the initial training data. Experimental results show that: 1) the proposed AGP is over 90% consistent with human gaze shifting path, as verified by the eye-tracking data, and 2) the retargeting algorithm outperforms its competitors significantly, as AGP is more indicative of photo esthetics than conventional saliency maps.

A Biologically Inspired Automatic System for Media Quality Assessment

Photo aesthetic quality evaluation is a challenging task in artificial intelligence systems. In this paper, we propose a biologically inspired aesthetic descriptor that mimicks humans sequentially perceiving visually/semantically salient In general, visually salient regions are perceived by low-level visual features, such as the high contrast between the foreground and the background objects; while semantically salient regions are perceived by high-level visual features such as human faces. regions in a photo. In particular, a weakly supervised learning paradigm is developed to project the local image descriptors into a low-dimensional semantic space. Then, each graphlet can be described by multiple types of visual features, both in low-level and in high-level. Since humans usually perceive only a few salient regions in a photo, a sparsity-constrained graphlet ranking algorithm is proposed that seamlessly integrates both the low-level and the high-level visual cues. Top-ranked graphlets are those visually/semantically prominent local aesthetic descriptors in a photo. They are sequentially linked into a path that simulates humans actively viewing process. Finally, we learn a probabilistic aesthetic measure based on such actively viewing paths (AVPs) from the training photos. Experimental results show that: 1) the AVPs are 87.65% consistent with real human gaze shifting paths, as verified by the eye-tracking data and 2) our aesthetic measure outperforms many of its competitors.

Retargeting Semantically-Rich Photos

Semantically-rich photos contain a rich variety of semantic objects (e.g., pedestrians and bicycles). Retargeting these photos is a challenging task since each semantic object has fixed geometric characteristics. Shrinking these objects simultaneously during retargeting is prone to distortion. In this paper, we propose to retarget semantically-rich photos by detecting photo semantics from image tags, which are predicted by a multi-label SVM. The key technique is a generative model termed latent stability discovery (LSD). It can robustly localize various semantic objects in a photo by making use of the predicted noisy image tags. Based on LSD, a feature fusion algorithm is proposed to detect salient regions at both the low-level and high-level. These salient regions are linked into a path sequentially to simulate human visual perception . Finally, we learn the prior distribution of such paths from aesthetically pleasing training photos. The prior enforces the path of a retargeted photo to be maximally similar to those from the training photos. In the experiment, we collect 217 $1600 \times 1200$ photos, each containing over seven salient objects. Comprehensive user studies demonstrate the competitiveness of our method.

Actively learning human gaze shifting paths for semantics-aware photo cropping.

Photo cropping is a widely used tool in printing industry, photography, and cinematography. Conventional cropping models suffer from the following three challenges. First, the deemphasized role of semantic contents that are many times more important than low-level features in photo aesthetics. Second, the absence of a sequential ordering in the existing models. In contrast, humans look at semantically important regions sequentially when viewing a photo. Third, the difficulty of leveraging inputs from multiple users. Experience from multiple users is particularly critical in cropping as photo assessment is quite a subjective task. To address these challenges, this paper proposes semantics-aware photo cropping, which crops a photo by simulating the process of humans sequentially perceiving semantically important regions of a photo. We first project the local features (graphlets in this paper) onto the semantic space, which is constructed based on the category information of the training photos. An efficient learning algorithm is then derived to sequentially select semantically representative graphlets of a photo, and the selecting process can be interpreted by a path, which simulates humans actively perceiving semantics in a photo. Furthermore, we learn a prior distribution of such active graphlet paths from training photos that are marked as aesthetically pleasing by multiple users. The learned priors enforce the corresponding active graphlet path of a test photo to be maximally similar to those from the training photos. Experimental results show that: 1) the active graphlet path accurately predicts human gaze shifting, and thus is more indicative for photo aesthetics than conventional saliency maps and 2) the cropped photos produced by our approach outperform its competitors in both qualitative and quantitative comparisons.

Weakly Supervised Photo Cropping

Photo cropping is widely used in the printing industry, photography, and cinematography. Conventional photo cropping methods suffer from three drawbacks: 1) the semantics used to describe photo aesthetics are determined by the experience of model designers and specific data sets, 2) image global configurations, an essential cue to capture photos aesthetics, are not well preserved in the cropped photo, and 3) multi-channel visual features from an image region contribute differently to human aesthetics, but state-of-the-art photo cropping methods cannot automatically weight them. Owing to the recent progress in image retrieval community, image-level semantics, i.e., photo labels obtained without much human supervision, can be efficiently and effectively acquired. Thus, we propose weakly supervised photo cropping, where a manifold embedding algorithm is developed to incorporate image-level semantics and image global configurations with graphlets, or, small-sized connected subgraph. After manifold embedding, a Bayesian Network (BN) is proposed. It incorporates the testing photo into the framework derived from the multi-channel post-embedding graphlets of the training data, the importance of which is determined automatically. Based on the BN, photo cropping can be casted as searching the candidate cropped photo that maximally preserves graphlets from the training photos, and the optimal cropping parameter is inferred by Gibbs sampling. Subjective evaluations demonstrate that: 1) our approach outperforms several representative photo cropping methods, including our previous cropping model that is guided by semantics-free graphlets, and 2) the visualized graphlets explicitly capture photo semantics and global spatial configurations.