The 3D extension of the High Efficiency Video Coding (3D-HEVC) standard has improved the coding efficiency for 3D videos significantly. However, this improvement has been achieved with a significant rise in computational complexity. Specifically, the encoding process for the depth map in the 3D-HEVC standard occupies 84% of the total encoding time. This extended time is primarily due to the need to traverse coding unit (CU) depth levels in depth map encoding to determine the most suitable CU size. Acknowledging the evident texture distribution patterns within a depth map and the strong correlation between encoding size selection and the texture complexity of the current encoding block, an adaptive depth early termination convolutional neural network, named ADET-CNN, is designed for the depth map in this paper. It takes an original 64 × 64 coding tree unit (CTU) as the input and provides segmentation probabilities for various CU sizes within the CTU, which eliminates the need for exhaustive calculations and the comparison for determining the optimal CU size, thereby enabling faster intra-coding for the depth map. Experimental results indicate that the proposed method achieves a time saving of 58% depth map encoding while maintaining the quality of synthetic views.

This study proposes a multi-view cupping spots image stitching method based on Canny SIFT to address issues such as limited field of view and unclear details of cupping spots during automatic cupping process. First, the image is preprocessed using bilateral filtering, and Canny edge detection is introduced to generate feature descriptors in combination with SIFT feature extraction algorithm. Then, RANSAC algorithm is utilized to screen and eliminate the mismatched feature points in order to obtain the optimal perspective transformation matrix. Finally, the weighted average fusion algorithm and morphological operations are combined to realize the multi-view cupping spots image splicing. Experimental results show that the algorithm proposed in this study improved the accuracy of feature matching for cupping spots images by 5% and 26% compared with the traditional SIFT and ORB algorithms, respectively. Moreover, the splicing speed also improved by 63%. The proposed method not only extracts more stable feature points, but also deals better with the problem of detailed features after fusion due to its improved splicing speed and quality.

Accurately predicting the remaining shelf life can effectively reduce the risk of spoilage during the storage process of agricultural products. The quality of agricultural products can be indirectly indicated by changes in environmental parameters. To better explore the intrinsic relationship between key environmental parameters during banana storage and their remaining shelf life, this paper proposes a novel causal convolution lightweight Transformer network. This model utilizes causal convolution operations to mine the temporal features of sensor data and applies positional encoding to the input signals. It employs a Transformer encoder to extract and fuse features while also utilizing a probabilistic sparse self-attention mechanism instead of the conventional self-attention mechanism. Moreover, a distillation operation is introduced, which effectively reduces the number of trainable parameters in the Transformer-based model and shortens the training time. Compared to traditional machine learning algorithms (BP, SVM) and conventional time series data mining algorithms (LSTM, RNN), the proposed prediction method achieves a mean squared error of 0.0221, a root mean squared error of 0.1486, a mean absolute error of 0.1101, and a maximum prediction error of 0.2221 days, allowing for more accurate and efficient predictions of bananas’ remaining shelf life.

Layout design is an important step in packaging graphic design, and high-quality layout design is an important attribute to attract consumers’ attention and subsequent purchase. In order to solve the problems of time-consuming, difficult communication and strong dependency faced by manually performing package graphic design, we propose a template-free package layout generation method to achieve intelligent layout design. This method uses generative adversarial network (GAN) as a framework, and the generator and discriminator are composed of two improved transformer structures, which combines the advantages of two mainstream generative models, taking into account the rich layout variations in packaging design for generating a robust layout. We also constructed a packaging dataset PackageLayout to verify the superiority of the proposed method, which contains 2020 packaging planar images and annotation information for three categories. After ablation experiments on the homemade packaging dataset and comparison experiments with current state-of-the-art methods (CGLGAN, DSGAN), we validated the effectiveness and stability of the model. The layouts generated by our model are visually similar to the real design layouts and outperform previous models in terms of evaluation metrics. Finally, we also constructed real designs based on the predicted layouts to better understand the visual quality, which contributes to the advancement of the application of intelligent layout design models in packaging graphic design.

The bearing vibration fault monitoring of advanced printing systems is crucial for ensuring system reliability, improving print quality, and enhancing production efficiency. In consideration of the high noise and strong interference attributes of bearing vibration signals of printing equipment caused by complex environmental factors, the original data noise interference is suppressed by a diagnostic approach for rolling bearing vibration signal faults combined Complementary Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN) and Dung Beetle Optimization-Support Vector Machine (DBO-SVM) proposed in this article. This method suppresses noise interference in the original data by decomposing and reconstructing vibration signals, and innovatively proposes the DBO-optimized SVM to address the problems of poor anti-interference ability and weak generalization ability of a single intelligent diagnostic method, effectively improving the fault diagnosis rate. First, the decomposition by CEEMDAN effectively overcomes the difficulties of modal aliasing and significant reconstruction errors found in traditional empirical mode decomposition (EMD). The optimal component reconstruction strategy considering the Correlation Coefficient and the Variance Contribution Rate is designed to obtain the vibration signal after noise reduction. Second, to effectively avoid the limitation of heavily relying on expert experience for hyperparameter adjustment, a DBO-SVM model is constructed utilizing a heuristic beetle optimization algorithm, dynamically optimizing the key kernel function parameters and penalty factors of SVM. Finally, the algorithm’s performance was tested using public datasets and self-tested data from Case Western Reserve University. The results indicate that the proposed approach achieves greater diagnostic accuracy and exhibits robust generalization.

In response to challenges such as large parameter count, difficulty in deployment, low accuracy, and slow speed of facial state recognition models in driver fatigue detection, the authors propose a lightweight real-time facial state recognition model called YOLOv5-fatigue based on YOLOv5n. First, a bilateral convolution is proposed, which can fully utilize the feature information in the channel. Then an innovative deep lightweight module is proposed, which reduces the number of network parameters as well as the computational effort by replacing the ordinary convolution in the neck network. Lastly, the normalization-based attention module is added to solve the problem of accuracy decline caused by lightweight models while keeping the number of parameters unchanged. In this paper, we first recognize the facial state by YOLOv5-fatigue and then use the proportion of eyes closed per unit of time and the proportion of mouth closed per unit of time to determine fatigue. In comparison experiments conducted on our self-built VIGP-fatigue dataset with other detection algorithms, our proposed method achieved an increase of 1% in AP50 compared to the baseline model YOLOv5n, reaching 92.6%. The inference time was reduced by 9% to 2.1 ms, and the parameter count decreased by 42.6% to 1.01 M.

Skin tumors have become one of the most common diseases worldwide. Usually, benign skin tumors are not harmful to human health, but malignant skin tumors are highly likely to develop into skin cancer, which is life-threatening. Dermoscopy is currently the most effective method of diagnosing skin tumors. However, the complexity of skin tumor cells makes doctors’ diagnoses subject to error. Therefore, it is essential to use computers for assisted diagnosis, thereby improving the diagnostic accuracy of skin tumors. In this paper, we propose Deep-skin, a model for dermoscopic image classification, which is based on both attention mechanism and ensemble learning. Considering the characteristics of dermoscopic images, we suggest embedding different attention mechanisms on top of Inception-V3 to obtain more potential features. We then improve the classification performance by late fusion of the different models. To demonstrate the effectiveness of Deep-skin, we conduct experiments and evaluations on the publicly available dataset Skin Cancer: Malignant vs. Benign and compare the performance of Deep-skin with other classification models. The experimental results indicate that Deep-skin performs well on the dataset in comparison to other models, achieving a maximum accuracy of 87.8%. In the future, we intend to investigate better classification models for automatic diagnosis of skin tumors. Such models can potentially assist physicians and patients in clinical settings. Link to dataset: Skin  Cancer:  Malignant  vs.  Benign  (kaggle.com)

Skin colour has a potential connection to facial impressions. For CG (computer graphics) character designing, tuning the skin colour to build a required character is a vital procedure. The previous findings provide essential suggestions to build an attractive face, but are still insufficient for the Artificial Intelligence (A.I.) system to build characters with different personalities and impressions. The present study investigates the relationship between skin colour and multiple categories of impression attributes based on a psychophysical experiment. Four attributes, which represent 23 attributes from personality and impression, were selected based on the principal component analysis. The relationship between these attributes and the hue angles was demonstrated which can be used to guide the AI system for character designing.