Search

A lightweight Mask R-CNN instance segmentation model was developed here to analyze particle size and shape accurately and quickly. Firstly, a hybrid Depthwise Dilated Convolutional Network (DDNet) is proposed, and the feature pyramid layers and the shared convolutional layers of the region proposal network are simplified, reducing the model complexity while ensuring robust feature extraction capabilities. Then, segmentation accuracy is significantly improved without sacrificing computational speed and performance by introducing the Dice loss function and clustering algorithm. Experimental results show that the model parameters are significantly reduced by 49.46%, and the segmentation speed increases from 2.15 FPS (frames per second) to 5.88 FPS. Meanwhile, the segmentation accuracy (AP50) increased from 90.56% to 91.21%. In addition, it was proven that the particle size distribution and shape could be analyzed accurately and rapidly with the proposed model, providing essential information for multiphase flow process optimization and equipment design in industrial applications. [Display omitted] • A lightweight Mask R-CNN instance segmentation model was proposed. • Adopting lightweight hybrid network increases segmentation speed markedly. • Loss function and candidate box significantly improved segmentation accuracy. • Particle size distribution and shape were measured accurately and rapidly. [ABSTRACT FROM AUTHOR]

In the context of the CERN EP R&D on monolithic sensors and the ALICE ITS3 upgrade, the Tower Partners Semiconductor Co (TPSCo) 65 nm process has been qualified for use in high energy physics, and adopted for the ALICE ITS3 upgrade. An Analog Pixel Test Structure (APTS) featuring fast per pixel operational-amplifier-based buffering for a small matrix of four by four pixels, with a sensor with a small collection electrode and a very non-uniform electric field, was designed to allow detailed characterization of the pixel performance in this technology. Several variants of this chip with different pixel designs have been characterized with a 120 GeV / c positive hadron beam. This result indicates that the APTS-OA prototype variants with the best performance achieve a time resolution of 63 ps with a detection efficiency exceeding 99% and a spatial resolution of 2 μm, highlighting the potential of TPSCo 65 nm CMOS imaging technology for high-energy physics and other fields requiring precise time measurement, high detection efficiency, and excellent spatial resolution. [ABSTRACT FROM AUTHOR]