Computer Vision, is a technology increasingly used in industrial processes. As we have already told you in previous posts, this technology is based on the acquisition of images or videos that are processed on a computer. Through this processing, very useful characteristics can be extracted for different applications, such as, for example: object classification, text reading, among others. Some basic characteristics that could be extracted would be, for example, the color or the outline of the objects in the analyzed image.
Computer vision with Deep Learning and Neural Networks
One of the techniques with the best results and most used in defect detection applications in combination with computer vision, is Deep Learning. This technique is a type of Artificial Intelligence based on the use of Deep Neural Networks, these networks are trained with previously taken images, which must be labeled to indicate what each image contains or what characteristics we are interested in analyzing. With these images, the network learns to extract the characteristics or information that is of interest to the application. This technique has become very popular due to the good results it obtains when extracting features from the image, allowing you to obtain results even with images that you have not been trained with.
Why Are Machine Vision Defect Detection Applications Necessary?
One of the main problems encountered in production processes is the appearance of defects. These defects may be due to various causes derived from the process itself and the conditions to which the treated material is exposed. The severity of these defects can be of different levels, which may cause the product created to be unfit or completely unusable due to said defect. For this reason the detection of these imperfections through the use of Computer Vision and Deep Learning techniques has become very important, since when detected in time and in an automated way they can be removed on time from the production line, or even, according to in which part of the process they are detected, to be corrected.
5 detection application with computer vision
- Detection of defects in metallic surfaces through compuer vision:
One of the most common cases in the detection of defects occurs on metallic objects or with metallic surfaces, in which it is very easy for several types of defects to occur in different phases of your production processes. The types of defects that can also appear are also many types and very different from each other, being able to be only of an aesthetic type such as stains, or much more damaging such as the appearance of notches, scratches, burns, lack of rectification, bumps, burrs, flatness, lack of thread, countersunk, rust or even cracks. This type of defect is also one of the most complex, due to the properties of the metal that makes it extremely sensitive to lighting, this being one of the most important elements when applying Computer Vision. For this reason it is essential to choose good lighting for these types of applications.
- Detection of Defects in fruits and vegetables through computer vision:
Another application that has gained great importance is the detection of defects for the food sector. Previously, this check was carried out manually by dedicated people, now thanks to advances in computer vision it is possible to automate this process, being faster and more efficient. This kind of defect detection ensures that the minimum quality requirements for fruits and vegetables are met and that they are not distributed with damage such as cuts, bumps or breaks. For this application, in many cases, hyperspectral cameras are used that allow distinguishing between the states of the materials, and, therefore, distinguishing between maturation phases, damage, etc.
- Detection of food defects through computer vision:
Computer vision is very useful in the food sector as it can be used in multiple applications and processes, such as the classification of different types of food. In particular, the detection of defects in this sector is of great importance, as it can be used to prevent the marketing of products that are contaminated or that could be harmful to health. In addition, it is possible to use Computer Vision to check the size of the food, or even to detect defects in the packaging itself to ensure the perfect preservation of the product.
- Detection of defects in assemblies through computer vision:
It is also possible to use computer vision to check the correct assembly of the different parts of an object. By means of computer vision it is possible to take images of the different components and check if they have been placed, if they are correctly placed, or if any of these components has its own defect. This type of application is gaining great importance in various sectors such as the automotive industry as it allows increasing the certainty that the elements produced will meet the required security.
- Detection of defects in electronic connections through computer vision:
Another application in which at ATRIA we use computer vision to detect defects is the verification of electronic connections. Using a camera and taking images, it can be verified that the connections of various electronic devices are correctly made, either by checking that none of the components are missing, or by checking that each of these elements is in the correct place and well connected. If a defect is detected in them, it can be quickly notified in various ways, or the repair could even be automated.
- Bonus track. Detection of defects in roads through computer vision:
Another application that is beginning to be used is the use of Computer Vision and Deep Learning to evaluate the condition of roads through images taken with drones. These images are taken from several meters high, so we can see a long stretch of road. Through these images it is possible to detect defects on the road surface and classify them according to the possible damage found and its severity. In this way, these damages can be located much faster, in addition to determining their location easily. Being able to detect them so quickly and easily allows them to be repaired long before they get worse and become defects that can impair road safety
Do you want to apply defect detection in any of your Projects? Contact us!