CellDefectNet: A Machine-designed Attention Condenser Network for Electroluminescence-based Photovoltaic Cell Defect Inspection

TL;DR

This paper introduces CellDefectNet, a lightweight, machine-designed neural network for efficient photovoltaic cell defect detection using electroluminescence images, achieving high accuracy with significantly reduced computational resources suitable for edge deployment.

Contribution

The work presents a novel attention condenser network, CellDefectNet, designed through machine-driven exploration for efficient defect detection on resource-constrained devices.

Findings

Achieves ~86.3% accuracy on a benchmark dataset.

Uses 13 times fewer parameters and FLOPs than EfficientNet-B0.

Runs 13 times faster on an ARM Cortex A-72 processor.

Abstract

Photovoltaic cells are electronic devices that convert light energy to electricity, forming the backbone of solar energy harvesting systems. An essential step in the manufacturing process for photovoltaic cells is visual quality inspection using electroluminescence imaging to identify defects such as cracks, finger interruptions, and broken cells. A big challenge faced by industry in photovoltaic cell visual inspection is the fact that it is currently done manually by human inspectors, which is extremely time consuming, laborious, and prone to human error. While deep learning approaches holds great potential to automating this inspection, the hardware resource-constrained manufacturing scenario makes it challenging for deploying complex deep neural network architectures. In this work, we introduce CellDefectNet, a highly efficient attention condenser network designed via machine-driven design exploration specifically for electroluminesence-based photovoltaic cell defect detection on the edge. We demonstrate the efficacy of CellDefectNet on a benchmark dataset comprising of a diversity of photovoltaic cells captured using electroluminescence imagery, achieving an accuracy of ~86.3% while possessing just 410K parameters (~13$\times$ lower than EfficientNet-B0, respectively) and ~115M FLOPs (~12$\times$ lower than EfficientNet-B0) and ~13$\times$ faster on an ARM Cortex A-72 embedded processor when compared to EfficientNet-B0.