Distracted driving is currently a global issue causing fatal traffic crashes and injuries. Although deep learning has achieved significant success in various fields, it still faces the trade-off between computation cost and overall accuracy in the field of distracted driving behavior recognition. This paper addresses this problem and proposes a novel lightweight attention-based (LWANet) network for image classification tasks. To reduce the computation cost and trainable parameters, we replace standard convolution layers with depthwise separable convolutions and optimize the classic VGG16 architecture by 98.16% trainable parameters reduction. Inspired by the attention mechanism in cognitive science, a lightweight inverted residual attention module (IRAM) is proposed to simulate human attention, extract more specific features, and improve the overall accuracy. LWANet achieved an accuracy of 99.37% on Statefarm’s dataset and 98.45% on American University in Cairo’s dataset. With only 1.22 M trainable parameters and a model file size of 4.68 MB, the quantitative experimental results demonstrate that the proposed LWANet obtains state-of-the-art overall performance in deep learning-based distracted driving behavior recognition.

Achieving accurate and high-sensitivity liquid level detection in medical instruments has always been a knotty task. In this paper, a high-precision, non-contact, flexible capacitive liquid level sensor is proposed, aiming to apply capacitive sensors in test tube liquid level measurement and improving the sensitivity of real-time liquid level sensors. The simulation study is conducted using ANSYS Maxwell and demonstrates the correlation between test tube thickness and sensitivity. A geometric model of the test container and sensing electrodes is established to optimize the design strategy for the physical dimensions of the sensor’s interdigitated (IDT) electrodes based on a flexible printed circuit (FPC). The hardware and software designs are completed based on the FDC2214 capacitive-to-digital converter to collect the capacitance variation data of the sensing electrodes accurately. To assess the system’s performance, an experimental platform for a liquid level sensor system has been constructed, facilitating the measurement, communication, processing, and visualization of liquid levels. The performance results demonstrate that the system is capable of accurately measuring the effective liquid level range within a standard 5 mL test tube with a resolution of up to 1 mm, as well as a sensitivity of 78.68 fF/mm, verifying the simulation results and exhibiting excellent linearity.