The changes in technology have led to a synchronous change in the clothing design method, as well as media and artistic aesthetics in the same period. The intelligence algorithm is constantly increasing its participation in development and production in the clothing industry. In this study, a variety of intelligent algorithms, including the parameterised number state algorithm, Generative Adversarial Networks, and style transfer were introduced into the multi-links of clothing product design and development, such as clothing shape, print pattern, texture craft, product vision, and so on. Men, an innovative clothing design method based on the cooperation of the intelligent algorithm and various human functional roles was constructed. The method improves the efficiency of the multiple links of clothing design, such as generating 10000 printing patterns every 72.12 seconds, and completing the migration of 92.7 frames of the garment process style every second. To a certain extent, this study realizes the scale economy of clothing design and reduces its marginal cost through the unlimited computing power brought about by Moore's law of digital technology, which provides a reference for the exploration of clothing design in the era of industry 4.0.

In the links of apparel product development and production, apparel pattern design cannot reduce its marginal cost through economies of scale because of its creative characteristics. With the world entering the era of industry 4.0, machine learning can provide services for apparel design. This research takes the Chinese characteristic tachisme pattern as the research object and puts forward a new design method of regional characteristic apparel pattern driven by Generative Adversarial Networks (GAN). Firstly, the main framework based on GAN including discrimination and generation modules is established. Aiming at the training difficulties of regional characteristic apparel pattern sample situation with small quantity and disordered specification, the image self-amplification and normalization pre-processing module is added to the model. Secondly, by adding the Batch Normalization mechanism, Leaky ReLU and RMSProp algorithm, the problems of gradient disappearance and overfitting in the experiment are solved, and the convergence speed of the model is improved. Finally, the HSV colour model algorithm is introduced into the loss function to indicate the training process, so that the artistic expression characteristics of the generated results are closer to the human visual perception experience. Through index evaluation comparison, result authenticity investigation and product design practice, we prove the superiority and practicality of the proposed method in this paper. The new design method theoretically solves the scale economy dilemma of the previous apparel pattern design methods and provides reference ideas for more application scenarios currently trapped in the real-time presentation of design results.

随着区域民族纺织品全球化格局的形成,中国风格成为了全球纺织品流行趋势之一.中国传统纺织图案在世界范围的流行对其设计开发方法提出了更高要求.文章以书法纺织图案为研究对象,提出了基于深度学习生成对抗网络的传统纺织图案开发方法.该方法解决了传统开发方法技术壁垒高、效率低、可控性差、资源损耗大等问题,同时在深度学习层面解决了中国传统纺织图案样本少、规格杂、重意不重形等训练难点.经对比实验、主观评估与设计运用,该方法相较当前典型方法更适合应用于传统纺织图案的设计开发,具备先进性与实践价值.

BackgroundThe efficient and accurate diagnosis of pulmonary adenocarcinoma before surgery is of considerable significance to clinicians. Although computed tomography (CT) examinations are widely used in practice, it is still challenging and time-consuming for radiologists to distinguish between different types of subcentimeter pulmonary nodules. Although there have been many deep learning algorithms proposed, their performance largely depends on vast amounts of data, which is difficult to collect in the medical imaging area. Therefore, we propose an automatic classification system for subcentimeter pulmonary adenocarcinoma, combining a convolutional neural network (CNN) and a generative adversarial network (GAN) to optimize clinical decision-making and to provide small dataset algorithm design ideas.MethodsA total of 206 nodules with postoperative pathological labels were analyzed. Among them were 30 adenocarcinomas in situ (AISs), 119 minimally invasive adenocarcinomas (MIAs), and 57 invasive adenocarcinomas (IACs). Our system consisted of two parts, a GAN-based image synthesis, and a CNN classification. First, several popular existing GAN techniques were employed to augment the datasets, and comprehensive experiments were conducted to evaluate the quality of the GAN synthesis. Additionally, our classification system processes were based on two-dimensional (2D) nodule-centered CT patches without the need of manual labeling information.ResultsFor GAN-based image synthesis, the visual Turing test showed that even radiologists could not tell the GAN-synthesized from the raw images (accuracy: primary radiologist 56%, senior radiologist 65%). For CNN classification, our progressive growing wGAN improved the performance of CNN most effectively (area under the curve =0.83). The experiments indicated that the proposed GAN augmentation method improved the classification accuracy by 23.5% (from 37.0% to 60.5%) and 7.3% (from 53.2% to 60.5%) in comparison with training methods using raw and common augmented images respectively. The performance of this combined GAN and CNN method (accuracy: 60.5%±2.6%) was comparable to the state-of-the-art methods, and our CNN was also more lightweight.ConclusionsThe experiments revealed that GAN synthesis techniques could effectively alleviate the problem of insufficient data in medical imaging. The proposed GAN plus CNN framework can be generalized for use in building other computer-aided detection (CADx) algorithms and thus assist in diagnosis.