The emergence of deep learning model GAN (Generative Adversarial Networks) is an important turning point in generative modeling. GAN is more powerful in feature and expression learning compared to machine learning-based generative model algorithms. Nowadays, it is also used to generate non-image data, such as voice and natural language. Typical technologies include BERT (Bidirectional Encoder Representations from Transformers), GPT-3 (Generative Pretrained Transformer-3), and MuseNet. GAN differs from the machine learning-based generative model and the objective function. Training is conducted by two networks: generator and discriminator. The generator converts random noise into a true-to-life image, whereas the discriminator distinguishes whether the input image is real or synthetic. As the training continues, the generator learns more sophisticated synthesis techniques, and the discriminator grows into a more accurate differentiator. GAN has problems, such as mode collapse, training instability, and lack of evaluation matrix, and many researchers have tried to solve these problems. For example, solutions such as one-sided label smoothing, instance normalization, and minibatch discrimination have been proposed. The field of application has also expanded. This paper provides an overview of GAN and application solutions for computer vision and artificial intelligence healthcare field researchers. The structure and principle of operation of GAN, the core models of GAN proposed to date, and the theory of GAN were analyzed. Application examples of GAN such as image classification and regression, image synthesis and inpainting, image-to-image translation, super-resolution and point registration were then presented. The discussion tackled GAN’s problems and solutions, and the future research direction was finally proposed.

The photovoltaic (PV) industry is an important part of the renewable energy industry. With the growing use of PV systems, interest in their operation and maintenance (O&M) is increasing. In this regard, analyses of power generation efficiency and inverter efficiency are very important. The first step in efficiency analysis is solar power estimation based on environment sensor data. In this study, solar power was estimated using a univariate linear regression model. The estimated solar power data were cross-validated with the actual solar power data obtained from the inverter. The results provide information on the power generation efficiency of the inverter. The linear estimation model developed in this study was validated using a single PV system. It is possible to apply the coefficients presented in this study to other PV systems, even though the nature and error rates of the collected data may vary depending on the inverter manufacturer. To apply the proposed model to PV systems with different power generation capacities, reconstructing the model according to the power generation capacity is necessary.

Insect identification has the disadvantage that it is difficult for non-experts to carry out due to the specificity of insects. Therefore, it is necessary for the general user to use auxiliary tools such as books to identify insects for education such as ecological learning. In recent years, researches using Deep Learning in fields such as object detection, behavior recognition, voice recognition as well as cancer detection in medical field have been actively conducted and show excellent results. In this paper, we developed a classification application that can be used in mobile phones with high automation and portability to solve the above insect classification problems. Experiments were conducted on 30 insect species selected for observable insects irrespective of environmental factors such as habitat and season, and the transform learning were applied to ResNet, which showed excellent performance in ILSVRC to classify forest insect. Our system achieved an average insect classification accuracy of 94%, an insect classification speed of 0.03 sec, and an insect photo transmission of 0.5 sec to output this information. This has proven that non-experts provide the appropriate performance to use.

In the field of deep learning, the generative model did not attract much attention until GANs (generative adversarial networks) appeared. In 2014, Google’s Ian Goodfellow proposed a generative model called GANs. GANs use different structures and objective functions from the existing generative model. For example, GANs use two neural networks: a generator that creates a realistic image, and a discriminator that distinguishes whether the input is real or synthetic. If there are no problems in the training process, GANs can generate images that are difficult even for experts to distinguish in terms of authenticity. Currently, GANs are the most researched subject in the field of computer vision, which deals with the technology of image style translation, synthesis, and generation, and various models have been unveiled. The issues raised are also improving one by one. In image synthesis, BEGAN (Boundary Equilibrium Generative Adversarial Network), which outperforms the previously announced GANs, learns the latent space of the image, while balancing the generator and discriminator. Nonetheless, BEGAN also has a mode collapse wherein the generator generates only a few images or a single one. Although BEGAN-CS (Boundary Equilibrium Generative Adversarial Network with Constrained Space), which was improved in terms of loss function, was introduced, it did not solve the mode collapse. The discriminator structure of BEGAN-CS is AE (AutoEncoder), which cannot create a particularly useful or structured latent space. Compression performance is not good either. In this paper, this characteristic of AE is considered to be related to the occurrence of mode collapse. Thus, we used VAE (Variational AutoEncoder), which added statistical techniques to AE. As a result of the experiment, the proposed model did not cause mode collapse but converged to a better state than BEGAN-CS.

Object detection automatically identifies and locates specific objects within images or videos, and plays a critical role in various fields. Among the many approaches, the YOLO model has significantly improved the accuracy and speed of object detection, thereby garnering attention for real-time processing. However, it exhibits limitations in detecting small or densely packed objects. This study introduces YOLO-RACE (You Only Look Once with ReAssembly and Convolutional block attEntion), an improved model based on YOLOv8 by incorporating the CARAFE upsampling technique, ResBlock, and the Convolutional Block Attention Module. YOLOv8 was selected from among various YOLO models because of its ability to deliver high accuracy in complex object detection tasks while maintaining its lightweight architecture and overall efficiency. The proposed model was evaluated using the VisDrone 2019, Pascal VOC, and SKU-110 K datasets. The experimental results demonstrated a precision of 0.419, recall of 0.321, mAP50 of 0.316, mAP50@95 of 0.18, and F1-score of 0.364 on the VisDrone 2019 dataset. On the Pascal VOC dataset, the model achieved a precision of 0.764, recall of 0.699, mAP50 of 0.773, mAP50@95 of 0.559, and F1-score of 0.73. Furthermore, on the SKU-110 K dataset, the model attained a precision of 0.904, recall of 0.841, mAP50 of 0.902, mAP50@95 of 0.575, and F1-score of 0.871. Based on these comparative results, the proposed model achieved superior performance compared with YOLOv8 and its variants, as well as with other state-of-the-art models such as YOLOv9, YOLOv10, YOLOv11, GELAN, and EfficientDet. This code is available at https://github.com/AnBLab-BAE/YOLO-RACE.git.

In the field of computer vision, convolutional neural network (CNN)-based models have demonstrated high accuracy and good generalization performance. However, in semantic segmentation, CNN-based models have a problem—the spatial and global context information is lost owing to a decrease in resolution during feature extraction. High-resolution networks (HRNets) can resolve this problem by keeping high-resolution processing layers parallel. However, information loss still occurs. Therefore, in this study, we propose an HRNet combined with an attention module to address the issue of information loss. The attention module is strategically placed immediately after each convolution to alleviate information loss by emphasizing the information retained at each stage. To achieve this, we employed a squeeze-and-excitation (SE) block as the attention module, which can seamlessly integrate into any model and enhance the performance without imposing significant parameter increases. It emphasizes the spatial and global context information by compressing and recalibrating features through global average pooling (GAP). A performance comparison between the existing HRNet model and the proposed model using various datasets show that the mean class-wise intersection over union (mIoU) and mean pixel accuracy (MeanACC) improved with the proposed model, however, there was a small increase in the number of parameters. With cityscapes dataset, MeanACC decreased by 0.1% with the proposed model compared to the baseline model, but mIoU increased by 0.5%. With the LIP dataset, the MeanACC and mIoU increased by 0.3% and 0.4%, respectively. The mIoU also decreased by 0.1% with the PASCAL Context dataset, whereas the MeanACC increased by 0.7%. Overall, the proposed model showed improved performance compared to the existing model.