react-native-vision-camera
react-native-vision-camera copied to clipboard
Using a photo set to test frame processor plugins❓
Question
How to test the accuracy of frames processor?
What I tried
I would like to be able to measure the accuracy of frame processor plugins QR code reader and OCR.
To do this I would like to be able to give to the two frame processor plugins a photo set and observe the results for each photo of the set and measure the accuracy ? To sum up, I would like to be able to test frame processor plugins alone with a photo set.
Ideas ? Thanks in advance :)
VisionCamera Version
latest
Additional information
- [ ] I am using Expo
- [X] I have read the Troubleshooting Guide
- [X] I agree to follow this project's Code of Conduct
- [X] I searched for similar questions in the issues page as well as in the discussions page and found none.
Well you can just count the successful reads yourself:
const successfulReads = useSharedValue(0)
const frameProcessor = useFrameProcessor((frame) => {
'worklet'
const textInImage = scanTextInImageOCR(frame)
if (textInImage.length > 0)
successfulReads.value += 1
}, [])
Thank you for your answer. However, this is not possible: To be able to measure the impact of image pre-processing (for example), we need to ALWAYS have the same inputs (exactly the same). This is the main problem. We really need a way to be able to control the input flow exactly.
Hi, I have similar use case where I'd like to pass an image to frameProcessor. Is it possible to map Image to Frame?
Did you find something ?
@Easy47 @wojteklembryk any updates on this? I am facing the same problem in which I want to be able to test the plugin using a static image.
@Easy47 @wojteklembryk any updates on this? I am facing the same problem in which I want to be able to test the plugin using a static image.
I made a frame processor plugin in which I load an image. I did not succeed in directly returning the frame so I test in the same plugin
@Easy47 @wojteklembryk any updates on this? I am facing the same problem in which I want to be able to test the plugin using a static image.
I made a frame processor plugin in which I load an image. I did not succeed in directly returning the frame so I test in the same plugin
I changed approach and instead of reusing frame processor I extracted it's logic into separate component.
My case was Google MLKit
based so I use image picker that's pass data to MLKit
bridge and then apply shared frame processor logic.
Closing as this is a stale issue - this might have been fixed with the full rewrite in VisionCamera V3 (🥳) - if not, please create a new issue.