Over the earlier time, a category of machinelearning, called deep learning, has attained significant achievements in several computer vision tasks such as image classification, object detection, semantic segmentation...
ISBN:
(数字)9798350348217
ISBN:
(纸本)9798350348224
Over the earlier time, a category of machinelearning, called deep learning, has attained significant achievements in several computer vision tasks such as image classification, object detection, semantic segmentation, patternrecognition and image classification generation. Deep learning objectives at finding various levels of dispersed representations, which have been proven to be discriminatively effective in many tasks. Distributed statement depicts similar information highlights across different adaptable and reliant layers. Each layer characterizes the data with a similar degree of exactness, however adapted to the degree of scale. The implementation of deep learning techniques depends greatly on the variety of data interpretation (or features) on which they are used. Artificial intelligence plans to understand interpretations of information regularly by changing over it or isolating components as of it, which creates it simpler to play out an undertaking like order or extrapolation.
For women, breast cancer ranks among the most fatal diseases. Mammography is the most widely used and effective imaging technique for the initial detection of breast cancer. On a mammography, the most significant indi...
详细信息
For women, breast cancer ranks among the most fatal diseases. Mammography is the most widely used and effective imaging technique for the initial detection of breast cancer. On a mammography, the most significant indicators of breast cancer are asymmetry, microcalcification, lumps, and distortion of the breast's architecture. What we call “microcalcification” (MC) refers to extremely minute deposits of calcium. It is common knowledge that the presence of clusters of microscopic calcification raises the cancer risk significantly. These microcalcification clusters are now being taken seriously as a reliable first signal in regular mammography screening. Microcalcifications occur in 61%-82% of malignant tumors, and in roughly 21% of these cases, they are the only indication of cancer. Microcalcifications on mammography film can be anywhere from 0.01mm2 to 1mm2. Considering how a radiologist's attention may be diverted from such minute calcium deposits. The diagnostic errors itself are the foundation of CAD in digital mammography. There are a plethora of computerized automated instruments for distinguishing and interpreting microcalcification's earlieststages. Diagnostic imaging, computer science, imageprocessing, patternrecognition, and AI are all brought together in computer assisted diagnosis. The detection performance and screening efficacy of mammography can be improved with the help of digital imageprocessing techniques, which CAD systems employ. CAD examines the digital mammography for suspicious regions and highlights them for the radiologist to further investigate. Once the suspicious area has been located, CAD separates it into sections, calculates many characteristics for each region, and uses the appropriate features to classify a lesion as benign or malignant.
In the 21st-century information age, with the development of big data technology, effectively extracting valuable information from massive data has become a key issue. Traditional data mining methods are inadequate wh...
详细信息
ISBN:
(数字)9798331534622
ISBN:
(纸本)9798331534639
In the 21st-century information age, with the development of big data technology, effectively extracting valuable information from massive data has become a key issue. Traditional data mining methods are inadequate when faced with large-scale, high-dimensional and complex data. Especially when labeled data is scarce, their performance is greatly limited. This study optimizes data mining algorithms by introducing semi-supervised learning methods, aiming to improve the algorithm's ability to utilize unlabeled data, thereby achieving more accurate data analysis and patternrecognition under limited labeled data conditions. Specifically, we adopt a self-training method and combine it with a convolutional neural network (CNN) for image feature extraction and classification, and continuously improve the model prediction performance through an iterative process. The experimental results demonstrate that the proposed method significantly outperforms traditional machinelearning techniques such as Support Vector machine (SVM), XGBoost, and Multi-Layer Perceptron (MLP) on the CIFAR-10 image classification dataset. Notable improvements were observed in key performance metrics, including accuracy, recall, and F1 score. Furthermore, the robustness and noise-resistance capabilities of the semi-supervised CNN model were validated through experiments under varying noise levels, confirming its practical applicability in real-world scenarios.
Problem-specific, well classified database is primary and most important requirement of all machinelearning-based systems. Hand-written character classification and recognition system is also no exception to this. In...
详细信息
Existing methods have been developed for light field (LF) image Super-Resolution (SR) and achieved continuously improved performance while suffering a significant performance drop when handling scenes with large dispa...
详细信息
ISBN:
(数字)9798350365474
ISBN:
(纸本)9798350365481
Existing methods have been developed for light field (LF) image Super-Resolution (SR) and achieved continuously improved performance while suffering a significant performance drop when handling scenes with large disparity variations. EPIT [1] was proposed to mitigate the disparity issue through non-local spatial-angular correlation learning. However EPIT has limitations due to the limited scale of existing LF datasets and the presence of imbalanced LF disparity, especially the scarcity of large disparity. To address this issue, we present a series of strategies to scale EPIT, called BigEPIT, including compound model scaling, augmented data resampling, and a high-precision test scheme. Specifically, the compound scaling method simultaneously scales the depth and width of the model to better improve the model capability. The augmented resampling method employs varying sampling intervals during training data generation, rather than solely relying on the central region view. This approach mitigates issues related to disparity imbalance and overfitting. The patch-based test scheme is popular because of its small GPU memory footprint. The traditional zero padding method and window partition will destroy the LF disparity structure and degrade the performance. Moreover, we find a positive correlation between the performance and the patchsize. Therefore, we advocate a high-precision test scheme i.e., a full-size or larger patchsize without zero padding for testing wherever the GPU memory permits, to achieve superior results. Extensive experiments demonstrate the effectiveness of our proposed method, which ranked 1st place in the NTIRE 2024 Light Field image Super-Resolution Challenge.
Matching features are integral to computer vision and can be used in image fusion, object recognition, or 3D model construction. Although conventional solutions are good, such as feature detection using SIFT with FLAN...
详细信息
ISBN:
(数字)9798331523657
ISBN:
(纸本)9798331523664
Matching features are integral to computer vision and can be used in image fusion, object recognition, or 3D model construction. Although conventional solutions are good, such as feature detection using SIFT with FLANN or brute-force and subsequent descriptor matching, they have severe problems when applying scale, rotation, or noise transformations. Other techniques, such as RANSAC, have also been used intensively to reduce match reliability by removing outliers. Nonetheless, the authors also discussed RANSAC when it has shortcomings; they stated that RANSAC has some drawbacks, such as being sensitive to parameter values, and when used with big data, it is very computationally expensive. To overcome these challenges, we suggest that SVMs be introduced into the feature-matching process in the SIFT framework. Matches are labeled inliers or outliers using detected training examples based on the SVM's descriptor similarity and geometrical consistency. This is an efficient replacement to the conventional LOS techniques because it includes probabilistic decisions, makes complex transformations plug-in with the methodology or the measurement system employed, as well as minimizing the critical dependence on the adjustment of the parameter values. A comparison of the performance of several theories shows that SVM-based filtering yields better match reliability, higher feature coverage, and a faster processing time than existing feature-matching approaches, and hence, it complements existing feature-matching solutions appropriately.
The use of active learning in supervised machinelearning is proposed in this study to reduce the expenses associated with labeling data. Active learning is a technique that includes iteratively selecting the most inf...
详细信息
A facial emotion recognition framework is proposed in this work. The convolutional neural network (CNN) has high ability in extraction of hierarchical spatial features from low level texture characteristics to high le...
A facial emotion recognition framework is proposed in this work. The convolutional neural network (CNN) has high ability in extraction of hierarchical spatial features from low level texture characteristics to high level contextual features. A simple CNN model with three layers is suggested in this paper where the features extracted in all layers containing multi-level features are activated and used for classification. In addition, the local binary pattern (LBP) descriptor is used to extract discriminative features from the spatial structure of the input image. Therefore, four feature sources are provided by multi levels of CNN and the LBP descriptor. Each feature source is used for facial emotion recognition by applying to the support vector machine (SVM) classier. Finally, the majority voting rule is used for decision fusion to provide the final emotional label of each given face image. The proposed method with 84% overall accuracy, 83% weighted F1-score and 81% kappa coefficient provides the best performance compared to LBP, multi-level CNN and two-dimensional principal component analysis (2DPCA) methods.
Automatic image captioning means the generation of a caption for an image by a machine. image captioning is performed by recognizing objects, attributes and interconnection between them. This task involves computer vi...
详细信息
Productivity in farming is very essential to the Indian economic system. When a plant gets any disease it can significantly reduce production, cost money and reduce the quality and quantity of farm produce. To prevent...
详细信息
暂无评论