Search Results (3)

In the process of steel strip production, the accuracy of defect detection remains a challenge due to the diversity of defect types, complex backgrounds, and noise interference. To improve the effectiveness of surface defect detection in steel strips, we propose an enhanced detection model known as YOLOv8-BSPB. First, we propose a novel pooling layer module, SCRD, which replaces max pooling with average pooling. This module introduces the receptive field block (RFB) and deformable convolutional network version 4 (DCNv4) to obtain learnable offsets, allowing convolutional kernels to flexibly move and deform on the input feature map, thus, more effectively extracting multi-scale features. Second, we integrate a polarized self-attention (PSA) mechanism to improve the model’s feature representation and enhance its ability to focus on relevant information. Additionally, we incorporate the BAM attention mechanism after the C2f module to strengthen the model’s feature selection capabilities. A bidirectional feature pyramid network is introduced at the neck of the model to improve feature transmission efficiency. Finally, the WIoU loss function is employed to accelerate the model’s convergence speed and enhance regression accuracy. Experimental results on the NEU-DET dataset demonstrate that the improved model achieves a classification accuracy of 81.3%, an increase of 4.9% over the baseline, with a mean average precision of 86.9%. The model has a parameter count of 5.5 M and operates at 103.1 FPS. To validate the model’s effectiveness, we conducted tests on the Kaggle steel strip dataset and our custom dataset, where the average accuracy improved by 2.3% and 5.5%, respectively. The experimental results indicate that the model meets the requirements for real-time, lightweight, and portable deployment. Full article

A new silicon-controlled rectifier embedded diode (SCR-D) for 7 nm bulk FinFET process electrostatic discharge (ESD) protection applications is proposed. The transmission line pulse (TLP) results show that the proposed device has a low turn-on voltage of 1.77 V. Compared with conventional SCR and diode string, the proposed SCR-D has an additional conduction path constituting by two additional inherent diodes, which results in a 1.8-to-2.2-times current surge capability as compared with the simple diode string and conventional SCR with the same size. The results show that the proposed device meets the 7 nm FinFET process ESD design window and has already been applied in actual circuits. Full article

This paper presents a switched capacitive reference driver (SCRD) with a low-energy switching scheme. In order to reduce the performance degradation resulting from a signal-dependent voltage drop in a capacitive reference driver (CRD) without increasing the capacitance (C_REF) of a CRD, the proposed SCRD utilizes the CRD for LSB conversion cycles. In MSB conversion cycles, a supply voltage is used as a reference voltage to save on area and power consumption. As such, the proposed SCRD significantly relaxes the required C_REF, and does not necessitate bit weight calibration or compensation requiring an auxiliary capacitor-based digital-to-analog converter (CDAC). To evaluate the proposed SCRD, a prototype 12-bit 40-MS/s SAR ADC is fabricated in a 65 nm CMOS process. With near Nyquist frequency, the measured spurious-free dynamic range (SFDR) of the SAR ADC with the SCRD is 80.6 dB, which is about a 16 dB improvement from the SFDR of a SAR ADC with a CRD only. Full article