To facilitate understanding of users' diverse queries against the back-end databases in web applications, researchers have introduced Text-to-SQL (Text2SQL) models that can generate well-structured SQL queries from users' query texts in natural language. As the Text2SQL model decouples the user queries with the back-end databases, it inherently mitigates the SQL injection risk posed by inserting users' input into pre-written SQL queries. However, what security risks to web applications may be posed by Text2SQL models remains an open question. In this paper, we present a new attack framework, named Autoregression-based Injection Attacks (AIA), to evaluate the security risks of Text2SQL models. In particular, AIA makes target models generate attack payloads by constructing specific inputs and adjusting the input auto-regressively. Our evaluation demonstrates that AIA can cause Text2SQL models to generate target output by adversarial inputs with success rates of over 70% in most scenarios. The generated adversarial input has certain transferability in target Text2SQL models. Additionally, practice experiments show that AIA can make Text2SQL models extract user lists from databases and even delete data in databases directly.

Scanning electron microscope image acquisition system is designed to control the scanning generator, collect the real-time image of the scanning electron microscope system, and process the image to help users get high quality and resolution sample images. Electronic microscope image acquisition system software which adapts Client/Server software architecture and develops based on the WPF framework, is aimed at controlling the scanning electron microscope imaging system. After further system testing, the result indicates that this system not only can realize the functions of image display in real-time, image measurement and annotation, image restoration and image storage, but also can attain high quality images by image restoration algorithms, which can meet automated and diversified requirements of the scanning electron microscope operators to capture electron microscope images.

The automatic segmentation of esophageal tissue layers in OCT images is essential for the study of esophageal diseases and computer-aided diagnosis. The tissue layer thickness of the esophagus is an important diagnostic sign for many esophageal diseases. Manually marking the boundary to calculate the average thickness of each layer is time-consuming and susceptible to subjective factors of the marker. In this paper, we propose a Pyramid Pooling Channel Attention Network (PPCANet) for the tissue segmentation. On the basis of the PSPNet, a channel attention module is introduced to selectively emphasize interdependent channel maps by integrating associated features among all channel maps, which makes the segmentation more precise. The potential clinical application of PPCANet for detecting eosinophilic esophagitis (EoE), an esophageal disease, is also presented in this paper.