Salmon inspection using Deep Learning

Generally, food and beverages need to be in a perfect condition when customers buy them. That is why this industry is very strict about the quality of their products. PEKAT VISION software can help with achieving this perfection. Thanks to the PEKAT Deep Learning software you can very precisely and also quickly detect defects.

In this specific case, Surface Detection module was used to inspect the salmon for bones and melanin spots.

There could be a lot of different defects you might look for when dealing with fish meat – e.g. bones, melanin spots, remaining blood, fish skin or scales. When searching for any of these defects in fish meat, the tricky part is that they are very hard to see and could take a significant amount of time to find, as you can see for yourself on the picture below:

But, with PEKAT VISION software you will be perfectly able to find all the defects you want to see very quickly:

The results of PEKAT VISION inspection are very precise and actually also easily achievable.

In this specific case we are using the Surface Detection module which is capable of performing a segmentation of defects and also their classification. We are looking for two types of defects – bones and melanin spots.

Firstly, we annotate all the defects and also classify them and then we start the training. We used 16 pictures for training in this case.

After training is finished, we can verify the results on images not included in the training.

After this, PEKAT VISION is ready to quickly and precisely detect all the defects. That is why AI is a perfect help and the right solution when you want to maintain or even improve the quality of almost any product.

Learn more about PEKAT Software