Search papers, labs, and topics across Lattice.
This study leverages transfer learning to enhance the detection and classification of Lithium-Ion Batteries (LIB) in X-ray images, focusing on three types: prismatic, pouch, and cylindrical. By training a large-scale dataset with a pre-trained YOLOv5m model, the authors achieved a precision of 94%, surpassing the baseline YOLOv5m performance by 5% while maintaining an efficient inference time of 22 ms. The findings underscore the effectiveness of transfer learning in specialized image classification tasks, particularly in the context of battery detection.
Achieving 94% precision in battery detection using transfer learning highlights a significant leap in image classification for critical applications.
The need for detecting and sorting batteries is drastically increasing for many applications. This study proves the potential of transfer learning in predicting whether the image contains a battery or not, the location and identifying three types of batteries, namely: prismatic, pouch, and cylindrical Lithium-Ion Batteries (LIB). Particularly, it focuses on the transfer learning method in two applications: Training a large-scale dataset to detect electronic devices using a pre-trained YOLOv5m, then using these latter trained weights to detect and classify the batteries. The precision of battery detection achieves 94%, which outperforms the pretrained YOLOv5m weights with 5%, in 22 ms inference time.
