Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

BatSort: Enhanced Battery Classification with Transfer Learning for Battery Sorting and Recycling

Yunyi Zhao, Wei Zhang, Erhai Hu, Qingyu Yan, Cheng Xiang, King Jet Tseng, Dusit Niyato

TL;DR

This work addresses the costly, largely manual sorting of used batteries and the data scarcity hindering ML applications in industrial recycling. It introduces BatSort, a transfer-learning framework that repurposes a ResNet-50V2 backbone by replacing its final layers to classify 9 battery types using a modest, in-house dataset (~500 images). Experimental results show an average accuracy of 92.1% with peaks at 96.2%, and a 2.03x improvement over a no-knowledge baseline. The approach enables fast, automated battery sorting with reduced labeling requirements and is extensible to other domains with limited data.

Abstract

Battery recycling is a critical process for minimizing environmental harm and resource waste for used batteries. However, it is challenging, largely because sorting batteries is costly and hardly automated to group batteries based on battery types. In this paper, we introduce a machine learning-based approach for battery-type classification and address the daunting problem of data scarcity for the application. We propose BatSort which applies transfer learning to utilize the existing knowledge optimized with large-scale datasets and customizes ResNet to be specialized for classifying battery types. We collected our in-house battery-type dataset of small-scale to guide the knowledge transfer as a case study and evaluate the system performance. We conducted an experimental study and the results show that BatSort can achieve outstanding accuracy of 92.1% on average and up to 96.2% and the performance is stable for battery-type classification. Our solution helps realize fast and automated battery sorting with minimized cost and can be transferred to related industry applications with insufficient data.

BatSort: Enhanced Battery Classification with Transfer Learning for Battery Sorting and Recycling

TL;DR

This work addresses the costly, largely manual sorting of used batteries and the data scarcity hindering ML applications in industrial recycling. It introduces BatSort, a transfer-learning framework that repurposes a ResNet-50V2 backbone by replacing its final layers to classify 9 battery types using a modest, in-house dataset (~500 images). Experimental results show an average accuracy of 92.1% with peaks at 96.2%, and a 2.03x improvement over a no-knowledge baseline. The approach enables fast, automated battery sorting with reduced labeling requirements and is extensible to other domains with limited data.

Abstract

Battery recycling is a critical process for minimizing environmental harm and resource waste for used batteries. However, it is challenging, largely because sorting batteries is costly and hardly automated to group batteries based on battery types. In this paper, we introduce a machine learning-based approach for battery-type classification and address the daunting problem of data scarcity for the application. We propose BatSort which applies transfer learning to utilize the existing knowledge optimized with large-scale datasets and customizes ResNet to be specialized for classifying battery types. We collected our in-house battery-type dataset of small-scale to guide the knowledge transfer as a case study and evaluate the system performance. We conducted an experimental study and the results show that BatSort can achieve outstanding accuracy of 92.1% on average and up to 96.2% and the performance is stable for battery-type classification. Our solution helps realize fast and automated battery sorting with minimized cost and can be transferred to related industry applications with insufficient data.
Paper Structure (22 sections, 2 equations, 3 figures, 2 tables)

This paper contains 22 sections, 2 equations, 3 figures, 2 tables.

Table of Contents

  1. Introduction
  2. System Architecture and Description
  3. Battery Inspection and Conveyor Belt
  4. Image-based Battery-type Classification
  5. Methodologies for BatSort
  6. Data Collection and Pre-processing
  7. Transfer Learning-based Battery-Type Classification
  8. Backbone Model
  9. Transferred Knowledge
  10. Backbone Reconfiguration
  11. Parameter Optimization
  12. Experimental Study
  13. Backbone Configuration and Experimental Setup
  14. Knowledge Transfer Performance
  15. Comparison to No Knowledge
  16. ...and 7 more sections

Figures (3)

  • Figure 1: An illustration of the system architecture of battery sorting. Battery type can be determined automatically with BatSort and the batteries of the same type are grouped into the same recycling bins for further processing.
  • Figure 2: An illustration of our transfer learning-based BatSort model with the last layers of ResNet-50V2 replaced with three new layers for customized classification and only the last seven layers of the new model trainable.
  • Figure 3: The BatSort's performance sensitivity in terms of accuracy to dropout rate, which varies between 0% to 50%, for both training and testing stages of the battery-type classification model. The red dashed line marks 95% accuracy for easy comparison. A high box means good accuracy and a small box implies stable performance. The optimal dropout rate with the best average accuracy and stability for battery-type classification is 20%.