3D Printing Failure Detection False Positives
What causes False Positives on your 3D Printer?
Although our AI model for detecting 3D Print Failures is extremely accurate and precise, it may occasionally detect some areas in the print bed as defects. Our AI model is trained on millions of data that are representative of most of the 3D printers out in the world. The data are comprised of images where failures are present and also where no failures are present. This allows for the AI model to learn what a failure is, and what it is not.
Sometimes the anything in a million item (perhaps your bonsai tree plant decoration :palm_tree:) in the background of the webcam frame looks like a spaghetti failure, and will be detected as one. This type of false positive can be caused by a few other common items such as:
- Ender 3 Logo on the print bed
- Leftover scraps near and on the print bed
- Highly-detailed infills (gyroid we're looking at you)
How can I minimize False Positives on my 3D Printer?
There are many steps you can take to reduce and eliminate the occurence of false positives on your 3d printer, including the following:
- Remove or cover the item causing the false positive. Cover the Ender logo with tape, remove the plant from being seen in your webcams frame, or change the orientation of the camera.
- Adjust the angle of your camera. This is specifically for false positives on the infill of the 3d printed part. Change the orientation of the camera so that is is even with the print bed and cannot see inside or over the print.
- Increase the sensitivity setting in the
PrintWatch
settings.
Additional Checks for your 3D Printers Camera
Here is a brief checklist for your 3D Printers camera to ensure the best format of image is going into the AI detection system:
- Ensure your camera is focused and not blurring
- Ensure the quality of your camera is AT LEAST 1280 x 720, with a recomended resolution of
1920 x 1080
- Ensure the
quality
parameter on your camera is set to 100% quality - Ensure the print bed fills a majority of the cameras frame from
left-to-right
andtop-to-bottom