Abstract
Tablet surface defect detection is an important part of tablet quality inspection, and manual detection and traditional pattern recognition methods are difficult to achieve the expected results. In this regard, this paper proposes a method for detecting tablet defects based on YOLOV3, which firstly uses industrial cameras to complete the collection of tablet defect images and creates a data set; then uses Daeknet-53 as the backbone network to initially extract features; secondly, constructs FPN feature pyramid enhancement Extraction of features; finally use yolo-head to obtain prediction results; detection mAP reaches 92.97% on the test set of the self-built data set. The experimental results show that the method has certain applicability and feasibility.
Highlights
IntroductionTablets go through a series of complex assembly line processes from production to packaging, usually including ingredients, sieving, mixing, granulation, drying, whole granulation, total mixing, tableting, coating, packaging and other steps [1]
The detection of surface defects of tablets in my country is mainly divided into manual detection and traditional machine vision detection
Manual detection often leads to eye fatigue due to long-term work in this method, the detection accuracy is reduced, and manual removal is laborintensive, low-efficiency, and prone to false detection and missed detection [2]; traditional machine vision The extraction effect of the technology on the feature is difficult to achieve the expected effect, and this method relies on the manual setting of the discriminant threshold, or on the manual selection of pattern features
Summary
Tablets go through a series of complex assembly line processes from production to packaging, usually including ingredients, sieving, mixing, granulation, drying, whole granulation, total mixing, tableting, coating, packaging and other steps [1]. The detection of surface defects of tablets in my country is mainly divided into manual detection and traditional machine vision detection. Deep learning detection algorithms are mainly divided into two categories in terms of ideas: one is the two-stage method, known as the region-based target detection algorithm. This type of method divides target detection into two parts: generating candidate frames and identifying target categories, such as Mask R-CNN[4], Faster R-CNN[5] and other algorithms, the advantage is that the detection accuracy is high. The YOLOV3 algorithm with balanced accuracy and real-time performance is used to detect tablet defects, and the trained model is tested using the test set, which has high accuracy
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
