Cuiling Jiang

Global contextual features are essential in computer vision tasks. Traditional convolutional networks are limited by the size of the convolutional kernel, resulting in a limited receptive field for each layer of the network. To address this issue, transformers introduced global attention, which has demonstrated excellent performance in natural language processing and has been widely applied in visual tasks. However, both convolutional networks and transformer models are constrained by the arrangement of data in Euclidean space, making it challenging to effectively extract features of irregular objects and complex scenes. We propose a graph neural network-based module (Graphion) for extracting global contextual features. Graphion maps the feature maps into a non-Euclidean space, establishes adjacency relationships among image patches, and uses a graph neural network for message passing and aggregation operations on the obtained node features, enabling the learning of correlation information from the graph structure. Consequently, it facilitates more efficient feature extraction of irregular object instances. Graphion is flexible and portable, allowing integration into visual feature extraction backbones. Extensive experiments were conducted on multiple datasets, including MS COCO, ADE20K, and ImageNet64. The proposed Graphion method demonstrates outstanding performance in object detection, segmentation, and classification.

Modern society requires increasing creative talents. As an educational concept, design thinking (DT) has been successfully applied in education to foster innovation. However, educational research on improving the engineering design ability of engineering students through the design process and practice is limited. Research has shown that project‐based learning (PBL) is an effective engineering education method. This research integrates DT and PBL (DT–PBL) in teaching the course “Introduction to Artificial Intelligence and Basic Algorithm Training” and discusses the teaching method and implementation of DT–PBL in detail through an example of a comprehensive training project of software and hardware collaboration based on a convolutional neural network and Raspberry Pi. This study considers the impact of DT–PBL on engineering students in terms of learning motivation and creativity using the pretest and posttest mixed method. Data analysis of quantitative tests and qualitative interviews shows that (1) The experimental group is superior to the comparison group in terms of overall learning motivation. Through a comprehensive training project of software and hardware collaboration, students can reflect on the required professional knowledge and skills, and improve their DT, engineering application, and innovation abilities and (2) The experimental group was superior to the comparison group in terms of overall creativity (adventure, curiosity, imagination, and challenge). Students in the experimental group can think more quickly and put forth more creative ideas from the perspective of user needs.

In view of the traditional teaching of the compulsory course "Data Structure", it is teacher-centered, without paying more attention to student learning outcomes. Outcomes-based education (OBE) is an educational model, student-centered instead of teacher-centered, and outcome-driven instead of content-driven. Considering the entire teaching process, including pre-class, in- class, and post- class, this paper discusses the course teaching reform approach and evaluation method in detail in terms of OBE. Through experiment design training from shallower to deeper, integration of theory and practice is cultivated, stimulating student interest and mastery in course study. The results demonstrate that OBE teaching improves the engineering application abilities of students, and the comprehensive quality of professional talents and the cultivation of diverse talents.

Interdisciplinary integration of theory and practice is imperative as a course requirement in emerging engineering education, and in the public elective course "Machine Vision Algorithm Training". Considering the entire teaching process, including pre-training, in-training, and post-training, this paper discusses the course construction and content in detail in terms of project-based learning (PBL). The PBL teaching approach and evaluation methods are described in detail through a comprehensive face recognition training case based on a convolutional neural network (CNN) and Raspberry Pi. Through project design training from shallower to deeper, interdisciplinary integration of theory and practice is cultivated, stimulating interest in course study. The results demonstrate that PBL teaching improves the engineering application and innovative abilities of students.

Image hash functions have wide-ranging application in many fields. This study presents a perceptual image-hashing scheme for a deep convolutional neural network (DCNN) for the purpose of content authentication. First, an AlexNet model of DCNN is constructed and trained to assess the performance of a given network. Then, the trained network is used to extract an image feature matrix. Finally, an image-hashing series is generated for content authentication. Experimental results show that, compared with other methods, the proposed method has a higher discrimination capability and an acceptable robustness against content-preserving operations, such as random attacks, rotation, JPEG compression, and additive Gaussian noise. A receiver operating characteristics curve is employed and demonstrates that the proposed image hashing obtains a desirable compromise between discrimination and robustness.

For the traditional steganographic method of Jsteg, the emmbedment of secret message is completed by dividing cover-image into nonoverlapping blocks of 8×8 pixels, discrete cosine transform (DCT) transforming, and using the standard 8×8 quantization table to quantize. In this paper, a novel steganographic method based on the JPEG quantization table modification is presented. Instead of dividing cover-image into 8×8 blocks, nonoverlapping blocks of 16×16 pixels is used. Both theoretical anlysis and experiment results show that the new methods has larger steganography capacity and better stego-image quality, compared with the method of Jsteg and Chang’s.

The paper puts forward a novel steganographic method based on the JPEG digital images. Instead of dividing cover-image into 8×8 blocks, non-overlapping blocks of 16×16 pixels is used. With our proposed quantization table, The DCT coefficients are quantized and embedded the secret messages. The experiment results show that our method has the larger steganography capacity and better stego-image quality than the other methods.