VoltaVision: A Transfer Learning model for electronic component classification

TL;DR

This work addresses electronic component recognition under limited labeled data. It introduces VoltaVision, a lightweight CNN designed for transfer learning from task-aligned pre-training data, and benchmarks it against large pretrained models. Results indicate that task-specific pre-training can match or surpass larger networks while offering smaller size and faster training, with robustness validated via 5-fold cross-validation on a 328-image, three-class dataset. The approach has practical implications for applications like e-waste recycling and automated component transactions, and the authors provide public code and dataset resources.

Abstract

In this paper, we analyze the effectiveness of transfer learning on classifying electronic components. Transfer learning reuses pre-trained models to save time and resources in building a robust classifier rather than learning from scratch. Our work introduces a lightweight CNN, coined as VoltaVision, and compares its performance against more complex models. We test the hypothesis that transferring knowledge from a similar task to our target domain yields better results than state-of-the-art models trained on general datasets. Our dataset and code for this work are available at https://github.com/AnasIshfaque/VoltaVision.

Figures (3)

• Figure 1: Experimental setups
• Figure 2: Overview of the electronic lending machine prototype
• Figure 3: Lending machine from user view and internal view