This guide is about conditions for image acceptance regarding data migration services supported by Machine Learning. It provides detailed explanations and diagrams regarding the quality, structure/appearance of image data to be compatible with our algorithms. Please note that we cannot successfully run our Machine Learning models against any such images.
-
- Multiple boxes that are not aligned
- Boxes have a gap between them (gap > than 1 row diameter height)
- Multiple boxes on different angles
- The model is designed to ignore incomplete core trays. If there are incomplete trays in the image that the client wants included in the crop, this will be a problem.
- Row detection is designed to find complete rows. If the row is cut off it is likely to be missed.
- If there are other complete trays in the image that the client wants to ignore, they will be included in the cropped image.
Too many boxes in one image
- Max rows is 20, but performance is most consistent when there are 1 or 2 core trays.