Abstract

Object detection in microelectronic circuits is critical for ensuring design integrity, manufacturing precision, and fault tolerance. With increasing circuit complexity and miniaturization, conventional imaging and detection approaches often fail to deliver the required accuracy and reliability. Existing object detection techniques struggle with high-resolution micro-scale structures, suffer from high false positives, and are computationally intensive. Moreover, integrating detection techniques within MEMS-based systems remains a challenge due to sensor limitations and noise sensitivity. This work proposes a SEMI MEMS (Smart Electro-Mechanical Integrated Micro System)- based supervised learning approach combining MEMS sensor data with convolutional neural networks (CNNs) for real-time object detection in microelectronic layouts. A custom-trained CNN is integrated with signal data from capacitive MEMS sensors to enhance feature extraction in noisy environments. The proposed method achieves 96.2% detection accuracy, a 15.3% improvement over baseline MEMS-CNN hybrids. Precision and recall values are 0.94 and 0.97, respectively. Compared to existing methods, processing time decreased by 22%, and false detection rate dropped by 18%.

Authors

Arvind Kumar Shukla¹, K. Regin Bose²
IFTM University, India¹, Rajalakshmi Institute of Technology, India²

Keywords

MEMS Sensors, Supervised Learning, Object Detection, Microelectronics, CNN