Detecting faulty bottle caps using CNN model

One of the most important stages in the development of any product is quality control. It is the process of determining the precise investigation of each sub-item of the product prior to arriving at the best client. Each stage of quality control is critical, especially in the food and drug industries, where, in addition to visual issues, extra security guidelines are required. Numerous quality control measures are frequently controlled by cameras. These cameras can be used for computer vision inspections as their significant benefits are quick execution, less consumption of power. The use of AI technologies in production lines could move the production lines even faster. This paper presents an image preprocessing strategy for detecting caps on bottles, which utilizes an added skipped connection between the blocks. Global Average Pooling is used to reduce the size of the parameters on caps. The proposed model shows much more accurate results than well recognized existing models , without loss of precision. So, for checking of the flawed caps, we have predicted using CNN to comprehend an identification method for detection of bottle caps in a production line that move at high rpm. In spite of some issues that are caused by taking an image from side, a classification issue which was found in bottles, class profound learning methods is deployed to solve this problem. Our model out-performs the notable frameworks that are right now utilized and it is additionally more productive than the popular models like VGG, Inception, and Res Net