Shelled nuts inspection and sorting
Identify acceptable nuts for placing on chocolates
関連製品
ディープラーニング採用の工業用画像解析向けグラフィカルプログラミング環境
ディープラーニング採用 In-Sight ViDi 画像処理ソフトウェア搭載
Shelled nuts on a conveyor belt need to be identified, picked up, and placed on top of chocolates with the correct side up by a pick-and-place robot. Edible nuts vary widely in shape, color, and texture, and have highly irregular boundaries. They are delicate and frequently end up on the conveyor broken, making them aesthetically unacceptable.
If the robot places a nut in the incorrect orientation or places a damaged or broken nut, the resulting chocolate will need to be rejected. If it is missed it may result in dissatisfied customers and brand damage.
The ranges of both possible acceptable shapes and unacceptable conditions of nuts are so wide, that it is impossible for conventional machine vision to reliably distinguish them, leading to unacceptably high error rates.
Cognex Deep Learning is ideal for the complex problem of finding undamaged nuts in the correct orientation, selecting them, and placing them on chocolates.
The classification tool trains on a set of images of acceptable, and damaged nuts. The tool rapidly classifies nuts into acceptable and unacceptable categories.
The part location tool trains on a set of images of acceptable nuts in the correct orientation and then locates them as they move past, no matter how many other nut pieces are also present.
Together these tools ensure that acceptable nuts are identified and located so that the pick and place arm can place them on the chocolates. Neither manual inspection nor other forms of machine vision can achieve anywhere near the same speed and accuracy in this task.
