Hello World,
I’m working NPU, which uses RTOS, even though I had found lot of libraries(Tesseract, EasyOCR and PaddleOCR) which can perform OCR, but I could not able to find out which one can be good for RTOS.
So can I know which OCR algorithm good for RTOS based NPUs?
I’d recommend looking at OpenCV DNN libraries.
Hello @ShaneKirkbride , What about Detection, which one is suitable for Edge detection?
And we planned to use DenseNet_CTC for recognition.
It would probably be overkill to use a dnn for edge detection if you are using a constrained system. There are a lot of good edge detection algorithms (sobel, laplace)which are small and fast. However, you may want to retrain DenseNet-CTC with the edge-filtered images for better results. But maybe I’m misunderstanding the workflow and goal.
Hello,
Can I use Paddle OCR to run on Edge Devices?
Depends on your goal and overall product requirements. The device would need t be able to run python since it looks like Paddle OCR is in python. This means it would consume more power, resulting in shorter battery life. For a quick prototype it is probably fine.
Hello I’m planning to use this https://arxiv.org/pdf/2009.09941 do you have any suggestions on it?
As of now the in Edge Device I have 128MBytes.