Journal
COMPUTER VISION, PT III
Volume 773, Issue -, Pages 178-186Publisher
SPRINGER-VERLAG SINGAPORE PTE LTD
DOI: 10.1007/978-981-10-7305-2_16
Keywords
Scratch detection; Product quality inspection; Cellphone screen; Convolutional neural networks
Categories
Funding
- National Key Research and Development Plan [2016YFC0801005]
- National Natural Science Foundation of China [61402463]
- Open Foundation Project of Robot Technology Used for Special Environment Key Laboratory of Sichuan Province in China [16kftk01]
Ask authors/readers for more resources
In the process of cellphone screen manufacture, equipment failures and human errors may lead to screen scratches. Traditional manual ways check scratches by human eyes, which often costs large manpower and time, but with poor effectiveness. To address this issue, we proposes an automated scratches detection method by cascading two main modules. First, the scratch filtering module detects big scratches and localizes small scratches candidates with a serial of low-level stages. Then, the scratches classification module applies a lightweight CNN model, ScratchNet, to identify each small scratch candidate whether it is real scratch or not. To train the ScratchNet, we build a Scratches on Cellphone Screen (SCS) dataset with 50K samples in 5 categories. The experimental results on SCS testing set show that the proposed method achieves an accuracy of 96.35% on classifying small scratches, which outperforms the LeNet model and the other classifiers.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available