1. We show that it is possible to accurately detect various types of defect in the complex microstructure of Li-ion battery from images of the electrodes using computer vision without the need for any hand-crafted feature extraction. 2.
Deep learning computer vision methods were used to evaluate the quality of lithium-ion battery electrode for automated detection of microstructural defects from light microscopy images of the sectioned cells.
Hence, the presented results demonstrate that deep learning architectures for battery detection and recognition in XRT images are a promising strategy for the cost-efficient detection and recognition of batteries in WEEE.
There is not much literature about defect detection in Li-ion battery electrode and to the best of our knowledge this is the first work to apply deep learning to this problem.
Taking leakage image for example, it comes in the form of an array of pixel values. The learned image feature in low-level layers represent the presence or absence of some elementary image features e.g. edge, texture and corner, at particular orientations and locations in the image.
During the stream of leakage recognition, pixel resolution of the test image is firstly resized into 500 × 500 from 3000 × 3000. The resized image is fed into leakage model which is acquired by FCN based on leakage dataset and segmented result of leakage will be output.