Quality control is a key process designed to ensure that only products satisfying the defined quality requirements reach the end consumer or the next step in a production line. In the food industry, in the packaging step, there are many products that are still evaluated by human operators. To automate the process and improve efficiency and effectiveness, computer vision and artificial intelligence techniques can be applied. This automation is challenging since specific strategies designed according to the application scenario are required. Focusing on the quality control of the sealing and closure of matrix-shaped thermoforming food packages, the aim of the article is to propose a deep-learning-based solution designed to automatically perform the quality control while satisfying production cadence and ensuring 100% inline inspection of the products. Particularly, the designed computer vision system and the image-based criteria defined to determine when a product has to be accepted or rejected are presented. In addition, the vision control software is described with special emphasis on the different convolutional neural network (CNN) architectures that have been considered (ResNet18, ResNet50, Vgg19 and DenseNet161, non-pre-trained and pre-trained on ImageNet) and on the specifically designed dataset. To test the solution, different experiments are carried out in the laboratory and also in a real scenario, concluding that the proposed CNN-based approach improves the efficiency and security of the quality control process. Optimal results are obtained with the pre-trained DenseNet161, achieving false positive rates that range from 0.03 to 0.30% and false negative rates that range from 0 to 0.07%, with a rejection rate between 0.64 and 5.09% of production, and being able to detect at least 99.93% of the sealing defects that occur in any production. The modular design of our solution as well as the provided description allow it to adapt to similar scenarios and to new deep-learning models to prevent the arrival of faulty products to end consumers by removing them from the automated production line.

Closure and seal inspection is one of the key steps in quality control of pizza packages. This is generally carried out by human operators that are not able to inspect all the packages due to cadence restrictions. To overcome this limitation, a computer vision system that automatically performs 100% inline seal and closure inspection is proposed. In this paper, after evaluating pizza package features, the manual quality control procedure, and the packaging machines of a real industrial scenario, a detailed description of hardware and software components of the proposed system as well as the main design decisions are presented. Focusing on the hardware, line-scan technology and hyperspectral imaging has been considered to ensure that all relevant information can be acquired independently of the pizza brand, topping, and film features. Focusing on the software, this applies a three-phases strategy that, first, applies a set of basic rejection controls; second, identifies the sealing region; and third, prepares the data for prediction through the classification of the pizzas using a deep learning network. This network is one of the software key elements and has been selected after comparing the commercial off-the-shelf (pretrained-dl-classifier-resnet50 from MVTec Halcon) and the custom-developed (ResNet18) architectures designed to automate the accept/reject classification of pizza packages. To train the networks, a classification of pizza package defects, focusing on sealing and closure, and an image-based method able to automatically detect them have been proposed. The system has been tested in laboratory and in real industrial conditions comparing it with the manual scenario and considering three pizza brands with two toppings per brand. From the evaluation, it has been seen that ResNet18 achieves the best results with mean, maximum, and minimum precision values of 99.87%, 99.95%, and 99.74%, respectively. Moreover, our system achieves twice the throughput rate with respect to the manual scenario, with the guarantee that all pizzas are evaluated, which is not possible in the manual scenario due to operator fatigue. The proposed solution can be easily adapted to similar contexts, even considering packages with other shapes.

No abstract