Sample Images Research Articles

A Novel High-Fidelity Simulation for Finishing Operations: Hybrid Image Mosaic and Wavelet Decomposition.

In finishing simulations, achieving accurate results can be challenging due to the minimal amount of material removal and the limited measurement range of surface micro-topography instruments. To overcome these limitations, a novel high-fidelity modeling method combining image mosaic and wavelet decomposition technologies is proposed in this paper. We achieve the stitching of narrow field and high pixel micro morphology images through four steps: image feature extraction, overlapped feature matching, feature fusion, and stitching effect evaluation. On this basis, the wavelet decomposition method is employed to separate detection signals based on their respective frequencies, allowing the establishment of a datum plane and a roughness surface. The point cloud model undergoes a transformation into a continuous geometric model via the Poisson reconstruction algorithm. In the case study, four sample images of an aluminum alloy sheet after barrel finishing were collected using the ZeGage Plus optical profiler. Each image has an actual size of 834.37 μm × 834.37 μm. Subsequently, a comparison was carried out between the physical and simulation experiments. The results clearly indicate that the proposed method has the potential to enhance the accuracy of the finishing simulation by over 30%. The error between the resulting model and the actual surface of the part can be controlled within 1 μm.

Analysis of Rock Mass Weathering Grade Using Image Processing Technique

Image processing techniques refer to the process of converting an image into a digital format and then performing various operations on it to extract useful information. In this study, image processing technique has been used to categorize rock masses according to its weathering grades. The pixel values of the sample images were in the form of RGB color space before being converted to CIELAB color space. The conversion uses ° as the illuminant. The value in CIELAB color space represents the green-red opponent colors, with negative values for green and positive values for red. In contrast, the value represents the blue-yellow opponents with negative values for blue and positive values for yellow. From the values of and of the samples, clustering was used to classify the samples. This method will group the and values into seven clusters according to the closest distance between the values and the centroids. The proposed study can differentiate between the rock mass and rejected clusters containing plants and painted numbers on the rock. The painted number is placed in a rejected cluster due to the inability to determine the exact color of the rock, thereby impacting the data accuracy. The results have been discussed, and the rock masses have been categorized based on weathering grade. Several limitations have been identified, such as the presence of shadows in the sample images and the lack of arrangement of outcome images according to their a* and b* values. This research has also been validated and compared with previous studies. The JudGeo software utilized in prior research required human input to manually estimate suitable a* and b* values, whereas the proposed method automatically computes these values during color space conversion of the sample. Additionally, the proposed method can calculate the percentage of each cluster, facilitating the classification of rock mass into its respective weathering grade.Image processing techniques refer to the process of converting an image into a digital format and then performing various operations on it to extract useful information. In this study, image processing technique has been used to categorize rock masses according to its weathering grades. The pixel values of the sample images were in the form of RGB color space before being converted to CIELAB color space. The conversion uses ° as the illuminant. The value in CIELAB color space represents the green-red opponent colors, with negative values for green and positive values for red. In contrast, the value represents the blue-yellow opponents with negative values for blue and positive values for yellow. From the values of and of the samples, clustering was used to classify the samples. This method will group the and values into seven clusters according to the closest distance between the values and the centroids. The proposed study can differentiate between the rock mass and rejected clusters containing plants and painted numbers on the rock. The painted number is placed in a rejected cluster due to the inability to determine the exact color of the rock, thereby impacting the data accuracy. The results have been discussed, and the rock masses have been categorized based on weathering grade. Several limitations have been identified, such as the presence of shadows in the sample images and the lack of arrangement of outcome images according to their a* and b* values. This research has also been validated and compared with previous studies. The JudGeo software utilized in prior research required human input to manually estimate suitable a* and b* values, whereas the proposed method automatically computes these values during color space conversion of the sample. Additionally, the proposed method can calculate the percentage of each cluster, facilitating the classification of rock mass into its respective weathering grade.

Fluorescence excitation-scanning hyperspectral imaging with scalable 2D–3D deep learning framework for colorectal cancer detection

Colorectal cancer is one of the top contributors to cancer-related deaths in the United States, with over 100,000 estimated cases in 2020 and over 50,000 deaths. The most common screening technique is minimally invasive colonoscopy using either reflected white light endoscopy or narrow-band imaging. However, current imaging modalities have only moderate sensitivity and specificity for lesion detection. We have developed a novel fluorescence excitation-scanning hyperspectral imaging (HSI) approach to sample image and spectroscopic data simultaneously on microscope and endoscope platforms for enhanced diagnostic potential. Unfortunately, fluorescence excitation-scanning HSI datasets pose major challenges for data processing, interpretability, and classification due to their high dimensionality. Here, we present an end-to-end scalable Artificial Intelligence (AI) framework built for classification of excitation-scanning HSI microscopy data that provides accurate image classification and interpretability of the AI decision-making process. The developed AI framework is able to perform real-time HSI classification with different speed/classification performance trade-offs by tailoring the dimensionality of the dataset, supporting different dimensions of deep learning models, and varying the architecture of deep learning models. We have also incorporated tools to visualize the exact location of the lesion detected by the AI decision-making process and to provide heatmap-based pixel-by-pixel interpretability. In addition, our deep learning framework provides wavelength-dependent impact as a heatmap, which allows visualization of the contributions of HSI wavelength bands during the AI decision-making process. This framework is well-suited for HSI microscope and endoscope platforms, where real-time analysis and visualization of classification results are required by clinicians.

DG2GAN: improving defect recognition performance with generated defect image sample

This article aims to improve the deep-learning-based surface defect recognition. In actual manufacturing processes, there are issues such as data imbalance, insufficient diversity, and poor quality of augmented data in the collected image data for product defect recognition. A novel defect generation method with multiple loss functions, DG2GAN is presented in this paper. This method employs cycle consistency loss to generate defect images from a large number of defect-free images, overcoming the issue of imbalanced original training data. DJS optimized discriminator loss is introduced in the added discriminator to encourage the generation of diverse defect images. Furthermore, to maintain diversity in generated images while improving image quality, a new DG2 adversarial loss is proposed with the aim of generating high-quality and diverse images. The experiments demonstrated that DG2GAN produces defect images of higher quality and greater diversity compared with other advanced generation methods. Using the DG2GAN method to augment defect data in the CrackForest and MVTec datasets, the defect recognition accuracy increased from 86.9 to 94.6%, and the precision improved from 59.8 to 80.2%. The experimental results show that using the proposed defect generation method can obtain sample images with high quality and diversity and employ this method for data augmentation significantly enhances surface defect recognition technology.

Characterization of Mass, Diameter, Density, and Surface Properties of Colloidal Nanoparticles Enabled by Charge Detection Mass Spectrometry.

A variety of scattering-based, microscopy-based, and mobility-based methods are frequently used to probe the size distributions of colloidal nanoparticles with transmission electron microscopy (TEM) often considered to be the "gold standard". Charge detection mass spectrometry (CDMS) is an alternative method for nanoparticle characterization that can rapidly measure the mass and charge of individual nanoparticle ions with high accuracy. Two low polydispersity, ∼100 nm diameter nanoparticle size standards with different compositions (polymethyl methacrylate/polystyrene copolymer and 100% polystyrene) were characterized using both TEM and CDMS to explore the merits and complementary aspects of both methods. Mass and diameter distributions are rapidly obtained from CDMS measurements of thousands of individual ions of known spherical shape, requiring less time than TEM sample preparation and image analysis. TEM image-to-image variations resulted in a ∼1-2 nm range in the determined mean diameters whereas the CDMS mass precision of ∼1% in these experiments leads to a diameter uncertainty of just 0.3 nm. For the 100% polystyrene nanoparticles with known density, the CDMS and TEM particle diameter distributions were in excellent agreement. For the copolymer nanoparticles with unknown density, the diameter from TEM measurements combined with the mass from CDMS measurements enabled an accurate measurement of nanoparticle density. Differing extents of charging for the two nanoparticle standards measured by CDMS show that charging is sensitive to nanoparticle surface properties. A mixture of the two samples was separated based on their different extents of charging despite having overlapping mass distributions centered at 341.5 and 331.0 MDa.

A ridge-based detection algorithm with filament overlap identification for 2D mycelium network analysis

Many terrestrial ecosystems engage mycorrhizal symbiotic associations, potentially to enhance nutrition, increase resistance to soil-borne pests and diseases, and improve resilience and soil structure. Mycorrhizal fungi create dynamic networked structures through branching and anastomosis that connect multiple plants and consent to transport resources underground from nutrient-rich patches to demanding plants. Controlled laboratory experiments are fundamental to improving our knowledge of mycelium network growth dynamics and further understanding its role in preserving ecological niches. We propose a method for highly automated analysis of the mycelium network structure and other morphological properties, such as hyphal length, hyphal density, and number of crossing and branches, in 2D microscopy images of fungal samples. Available tools for automated network analyses suffer from overestimating network connectivity since filament crossings are not considered. In particular, we propose a) a ridge-based mycelium detection algorithm and b) a geometrical-based approach to identify overlapping filaments crossing each other. The algorithmic solution is evaluated on a total of 135 real mycelium sample images over different validation steps, originating from different datasets and having different characteristics, including background, contrast, image acquisition system, fungal species, and clearness (e.g., level of transparency, homogeneity, dirtiness of the medium) of the sample. Results show that 1) the proposed detection method can be used to measure the length of mycelium in an image, replacing manual tracing and allowing for less laborious analysis ρ̂c=0.96, 2) the filament detection is on par with state-of-the-art techniques F1=0.88−0.94 with a more intuitive parameterization, and 3) the proposed algorithm correctly identifies filament crossings F1=0.89 in most common cases, yielding a reduction in the overestimation of network connectivity. The latter feature consents to applying the proposed fully automated solution to complex and irregular fungal structures, advancing mycelium detection and reconstruction performance accuracy with respect to the state-of-the-art.

Intelligent identification system of wild animals image based on deep learning in biodiversity conservation law

This study aims to overcome the impact of complex environmental backgrounds on the recognition of wildlife in monitoring images, thereby exploring the role of a deep learning-based intelligent wildlife recognition system in biodiversity conservation. The automatic identification of wildlife images is conducted based on convolutional neural networks (CNNs). Target detection technology, based on regression algorithms, is initially employed to extract Regions of Interest (ROI) containing wildlife from images. The wildlife regions in monitoring images are detected, segmented, and converted into ROI images. A dual-channel network model based on Visual Geometry Group 16 (VGG16) is implemented to extract features from sample images. Finally, these features are input into a classifier to achieve wildlife recognition. The proposed optimized model demonstrates superior recognition performance for five wildlife species, caribou, lynx, mule deer, badger, and antelope, compared to the dual-channel network model based on VGG16. The optimized model achieves a Mean Average Precision (MAP) of 0.714, with a maximum difference of 0.145 compared to the other three network structures, affirming its effectiveness in enhancing the accuracy of automatic wildlife recognition. The model effectively addresses the issue of low recognition accuracy caused by the complexity of background information in monitoring images, achieving high-precision recognition and holding significant implications for the implementation of biodiversity conservation laws.

Through the lens of color: Relationships between color properties and visual perception of wood (Acer.) figures

The wood figure is a crucial wood trait that contributes to its versatile applications, drawing attention and inspiring imagination. This study aims to elucidate the impact of color attributes on the visual perception of wood figures. Initially, using intelligent methods (interactive color palettes), we colored six typical figures of natural wood (Acer.) as sample images. These images were characterized by entropy values, and assessed by simulated visual saliency (ITTI model) and subjective questionnaires. Finally, we discussed how different color quantities and schemes affect complexity and visual saliency of wood figures. The results indicate that changes in color attributes significantly impact the visual perception of wood figures. When the number of replacement colors is four, both the change rate in image complexity and the mean visual saliency are at their highest, at 24% and 0.335 respectively. Among the colors red, yellow, and green, red tends to influence the intensity of visual saliency more, the subjective complexity of green is higher. In addition, we found that more complex figures are less affected by changes in color attributes, and people can also distinguish the degree of difference between figure types well.

A Super Encryption Approach for Enhancing Digital Security using Column Transposition - Hill Cipher for 3D Image Protection

Image encryption is an indispensable technique in the realm of information security, serving as a pivotal mechanism to safeguard visual data against unauthorized access and potential breaches. This study scrutinizes the effectiveness of merging columnar transposition with the Hill Cipher methodologies, unveiling specific metrics from a curated set of sample images. Notably, employing column transposition with the key "JAYA" and the Hill Cipher with the key "UDINUSSMG," the encrypted images underwent rigorous evaluation. 'Lena.png' demonstrated an MSE of 513.32 with a PSNR of 7.89 dB, while 'Peppers.png' and 'Baboon.png' recorded MSE values of 466.67 and 423.92, respectively, with corresponding PSNR figures of 7.12 dB and 7.31 dB. Across all samples, a consistent BER of 50.00% indicated uniform error propagation, while entropy values settled uniformly at 7.9999, highlighting consistent data complexity. While the findings underscore a consistent error rate and complexity, there's a compelling need for further refinement to enhance image quality and security. Moreover, the study proposes future research avenues exploring a three-layer super encryption paradigm, amalgamating columnar transposition, Hill Cipher, and other robust algorithms. This approach aims to fortify encryption methodologies against evolving threats and challenges in data protection, offering heightened resilience and efficacy in safeguarding sensitive information.

Utilizing Deep Learning for Defect Inspection in Hand Tool Assembly.

The integrity of product assembly in the precision assembly industry significantly influences the quality of the final products. During the assembly process, products may acquire assembly defects due to personnel oversight. A severe assembly defect could impair the product's normal function and potentially cause loss of life or property for the user. For workpiece defect inspection, there is limited discussion on the simultaneous detection of the primary kinds of assembly anomaly (missing parts, misplaced parts, foreign objects, and extra parts). However, these assembly anomalies account for most customer complaints in the traditional hand tool industry. This is because no equipment can comprehensively inspect major assembly defects, and inspections rely solely on professionals using simple tools and their own experience. Thus, this study proposes an automated visual inspection system to achieve defect inspection in hand tool assembly. This study samples the work-in-process from three assembly stations in the ratchet wrench assembly process; an investigation of 28 common assembly defect types is presented, covering the 4 kinds of assembly anomaly in the assembly operation; also, this study captures sample images of various assembly defects for the experiments. First, the captured images are filtered to eliminate surface reflection noise from the workpiece; then, a circular mask is given at the assembly position to extract the ROI area; next, the filtered ROI images are used to create a defect-type label set using manual annotation; after this, the R-CNN series network models are applied to object feature extraction and classification; finally, they are compared with other object detection models to identify which inspection model has the better performance. The experimental results show that, if each station uses the best model for defect inspection, it can effectively detect and classify defects. The average defect detection rate (1-β) of each station is 92.64%, the average misjudgment rate (α) is 6.68%, and the average correct classification rate (CR) is 88.03%.

Is human perception reliable? Toward illumination robust food freshness prediction from food appearance — Taking lettuce freshness evaluation as an example

Machine vision-based food quality evaluation systems have achieved great attention in industry and academia in recent years because of their high-throughput and non-invasive properties. In practice, the environmental illumination conditions variations would cause visual evaluation bias for both human and machine perceptions. Compared to other existing studies which take sample images and human visual grading under fixed illumination conditions, this study first investigated the environmental illumination effects on the performance of visual-based food grading for both humans and machines. Taking lettuce samples as an example, an image dataset that considered the environmental illumination variations was first established. This dataset encompasses human visual grading scores obtained from sensory panels as well. In contrast to current studies that utilize a single grader or L*a*b* color features to assess sample freshness, our study reveals a substantial impact of environmental illumination on both human perception (p<0.0001) and L*a*b* color features (p<0.0001). In order to enhance the performance of machine vision-based freshness prediction, multitask learning protocols were incorporated into the proposed new network architectures. This allowed the simultaneous prediction of both sample freshness and illumination conditions. In comparison to commonly used generic convolutional neural network models and vision transformer models, the newly proposed model exhibited superior freshness prediction performance. It minimized the prediction error by 20.36%, outperforming the generic ResNet model. This research represents the first quantitative study addressing human and machine perceptions under varied illumination conditions for food quality evaluation. The findings are anticipated to play a pivotal role in expediting the integration of machine vision applications into food engineering practices.

Real-Time Black Ice Detection in Drone View Using YOLOX

Black ice is a perilous phenomenon resulting from frozen precipitation on road surfaces. It stands as one of the most prominent contributors to winter traffic accidents since it is nearly invisible. Given these characteristics, there is a pressing need to develop automotive systems capable of detecting black ice to improve driving safety and conditions. This study proposes the deployment of two YOLOX-based models to implement real-time black ice detection from drone-view images. The proposed approach leverages YOLOX-Tiny, known for its proficiency in real-time object detection, in conjunction with YOLOX-B which is a refined version designed specifically for detecting black ice on roadways. For training and evaluating the proposed system, we collected a dataset of black ice consisting of 2,851 sample images sized at 416x416 pixels. Significantly, YOLOX-Tiny achieved an AP@[0.5:0.95] of 0.4923, whereas YOLOX-B achieved 0.4779. Additionally, we demonstrate a practical implementation of the proposed method by deploying the system on the Nvidia Jetson Orin Nano device for real-time inference. The black ice dataset we compiled is publicly available on GitHub. We expect that the proposed black ice detection system will contribute to effectively maintaining safer driving conditions.

A comprehensive evaluation method for dust pollution: Digital image processing and deep learning approach

Dust pollution poses a grave threat to both the environment and human health, especially in mining operations. To combat this issue, a novel evaluation method is proposed, integrating grayscale average (GA) analysis and deep learning (DL) in image classification. By utilizing a self-designed dust diffusion simulation system, 300 sample images were generated for analysis. The GA method establishes a correlation between grayscale average and dust mass, while incorporating fractal dimension (FD) enhances classification criteria. Both GA and DL methods were trained and compared, yielding promising results with a testing accuracy of 92.2 % and high precision, recall, and F1-score values. This approach not only demonstrates efficacy in classifying dust pollution but also presents a versatile solution applicable beyond mining to diverse dust-contaminated work environments. By combining image processing and deep learning, it offers an automated and reliable system for environmental monitoring, thereby enhancing safety standards and health outcomes in affected industries. Ultimately, this innovative method signifies a significant advancement towards mitigating dust pollution and ensuring sustainable industrial practices.

Testing the generalizability and effectiveness of deep learning models among clinics: sperm detection as a pilot study

BackgroundDeep learning has been increasingly investigated for assisting clinical in vitro fertilization (IVF). The first technical step in many tasks is to visually detect and locate sperm, oocytes, and embryos in images. For clinical deployment of such deep learning models, different clinics use different image acquisition hardware and different sample preprocessing protocols, raising the concern over whether the reported accuracy of a deep learning model by one clinic could be reproduced in another clinic. Here we aim to investigate the effect of each imaging factor on the generalizability of object detection models, using sperm analysis as a pilot example.MethodsAblation studies were performed using state-of-the-art models for detecting human sperm to quantitatively assess how model precision (false-positive detection) and recall (missed detection) were affected by imaging magnification, imaging mode, and sample preprocessing protocols. The results led to the hypothesis that the richness of image acquisition conditions in a training dataset deterministically affects model generalizability. The hypothesis was tested by first enriching the training dataset with a wide range of imaging conditions, then validated through internal blind tests on new samples and external multi-center clinical validations.ResultsAblation experiments revealed that removing subsets of data from the training dataset significantly reduced model precision. Removing raw sample images from the training dataset caused the largest drop in model precision, whereas removing 20x images caused the largest drop in model recall. by incorporating different imaging and sample preprocessing conditions into a rich training dataset, the model achieved an intraclass correlation coefficient (ICC) of 0.97 (95% CI: 0.94-0.99) for precision, and an ICC of 0.97 (95% CI: 0.93-0.99) for recall. Multi-center clinical validation showed no significant differences in model precision or recall across different clinics and applications.ConclusionsThe results validated the hypothesis that the richness of data in the training dataset is a key factor impacting model generalizability. These findings highlight the importance of diversity in a training dataset for model evaluation and suggest that future deep learning models in andrology and reproductive medicine should incorporate comprehensive feature sets for enhanced generalizability across clinics.

Imaging performance of portable and conventional ultrasound imaging technologies for ophthalmic applications.

The aim of this study was to evaluate the imaging capabilities of Butterfly iQ with conventional ophthalmic (piezoelectric) ultrasound (COU) for ophthalmic imaging. Custom phantom molds were designed and imaged with Butterfly iQ and COU to compare spatial resolution capabilities. To evaluate the clinical imaging performance of Butterfly iQ and COU, a survey containing pathological conditions from human subjects, imaged with both Butterfly iQ and COU probes, was given to three retina specialists and graded on image detail, resolution, quality, and diagnostic confidence on a ten-point Likert scale. Kruskal-Wallis analysis was performed for survey responses. Butterfly iQ and COU had comparable capabilities for imaging small axial and lateral phantom features (down to 0.1 mm) of high and low acoustic reflectivity. One of three retina specialists demonstrated a statistically significant preference for COU related to resolution, detail, and diagnostic confidence, but the remaining graders showed no significant preference for Butterfly iQ or COU across all sample images presented. The emergence of portable ultrasound probes offers an affordable alternative to COU technologies with comparable qualitative imaging resolution down to 0.1 mm. These findings suggest the value to further study the use of portable ultrasound systems and their utility in routine eye care.

Full inference for the anisotropic fractional Brownian field

The anisotropic fractional Brownian field (AFBF) is a non-stationary Gaussian random field which has been used for the modeling of textured images. In this paper, we address the open issue of estimating the functional parameters of this field, namely the topothesy and Hurst functions. We propose an original method which fits the empirical semi-variogram of an image to the semi-variogram of a turning-band field that approximates the AFBF. Expressing the fitting criterion in terms of a separable non-linear least square criterion, we design a minimization algorithm inspired by the variable projection approach. This algorithm also includes a coarse-to-fine multigrid strategy based on approximations of functional parameters. Compared to existing methods, the new method enables to estimate both functional parameters on their whole definition domain. On simulated textures, we show that it has a low estimation error, even when the parameters are approximated with a high precision. We also apply the method to characterize mammograms and sample images with synthetic parenchymal patterns.

Lensless shadow microscopy-based shortcut analysis strategy for fast quantification of microplastic fibers released to water

Fast quantification is the primary challenge in monitoring microplastic fiber (MPF) pollution in water. The process of quantifying the number of MPFs in water typically involves filtration, imaging on a filter membrane, and manual counting. However, this routine workflow has limitations in terms of speed and accuracy. Here, we present an alternative analysis strategy based on our high-resolution lensless shadow microscope (LSM) for rapid imaging of MPFs on a chip and modified deep learning algorithms for automatic counting. Our LSM system was equipped with wide field-of-view submicron-pixel imaging sensors (>1 cm2; ∼500 nm/pixel) and could simultaneously capture the projection image of >3-μm microplastic spheres within 90 s. The algorithms enabled accurate classification and detection of the number and length of >10-μm linear and branched MPFs derived from melamine cleaning sponges in each image (∼0.4 gigapixels) within 60 s. Importantly, neither MPF morphology (dispersed or aggregated) nor environmental matrix had a notable impact on the automatic recognition of the MPFs by the algorithms. This new strategy had a detection limit of 10 particles/mL and significantly reduced the time of MPF imaging and counting from several hours with membrane-based methods to just a few minutes per sample. The strategy could be employed to monitor water pollution caused by microplastics if an efficient sample separation and a comprehensive sample image database were available.

Infrared small target detection algorithm based on filter kernel combination optimization learning method

Infrared small target detection in complex backgrounds remains a big challenge due to its small size and weak energy, which makes it easy to be submerged in complex backgrounds. Common filtering algorithms are difficult to apply to various complex scenes and generate many false alarms and poor robustness. Therefore, this paper proposes an infrared small target detection algorithm based on filter kernel combination optimization learning method. For the sample image block use typical filter kernels and weight coefficients to calculate the combination of filtering results, and construct the loss function of the results and the true value. The optimal weight coefficients are learned using an optimization method, which effectively achieves the fusion of multiple features, then we build a sample library containing comprehensive priori information. In the implementation, an online inference method based on similarity measure is proposed to match test images with sample library image blocks to obtain optimal combination weights. By utilizing the comprehensive priori information to enhance the adaptive ability of the algorithm for various complex backgrounds. Finally, get the position of the infrared small target. The experimental results show that our algorithm has excellent detection performance for infrared small targets with complex background and low signal to clutter ratio, and the AUC value is 0.9980, which is more robust than other comparative algorithms.

Image Caption Generator using Deep Learning

Image Caption Generation has always been a study of great interest to the researchers in the Artificial Intelligence department. Being able to program a machine to accurately describe an image or an environment like an average human has major applications in the field of robotic vision, business and many more. Automatic caption generation with attention mechanisms aims at generating more descriptive captions containing coarse to fine semantic contents in the image. This has been a challenging task in the field of artificial intelligence. In this paper, we present different image caption generating models based on deep neural networks, focusing on the various CNN techniques and analyzing their influence on the sentence generation. We have also generated captions for sample images and compared the different feature extraction and encoder models to analyse which model gives better accuracy and generates the desired results

The Combination of Feature Extraction and Classification by Bag of Visual Words to Detect Breast Cancer for Improved Accuracy

The main objective is to improve diagnostic accuracy of the breast cancer. The goal is to improve the accuracy of breast cancer detection by leveraging both the discriminative power of feature extraction and the robustness of classification using BoVW. By combining these techniques, the system aims to effectively distinguish between cancerous and non-cancerous tissues in medical images, thereby aiding in early diagnosis and treatment planning for breast cancer patient. The breast cancer sample images were gathered from the city's hospitals Chennai 2023 as well as the mini-MIAS database and DDSM database. The BoVW classifier is used to identify and classify breast tumor images according to their size. This improves the efficacy of the classification method developed for the CAD system's mammography breast cancer picture categorization. Several Computer Aided Diagnosis (CAD) systems have been created to use mammography images to identify breast cancer in its early stages. ANN attained 93.6%,94.18%, 93.2% and SVM gained 92%,92.44%,93.02% for Precision, Recall and Accuracy Respectively.The proposed method display excellent accuracy of 98.83% over the other methods. The novelty could stem from the seamless integration of multiple modalities within the feature extraction and classification framework.