Could you help me suggest one development board?

[Lchen.11]

1. base on STM32 and AI platform
2. Handwritten number recognition
3. ​the recognized number can be voice broadcast, for example 1.
4. the handwritten speed can be record by camera.

[Panometric]

You can use STM32Cube to filter down possible hardware. Seems you need a DCMI (Camera) and an SAI (Audio) but there are still hundreds of choices. From there I would look for ones with development boards and the fastest / most memory to experiment on. Optimize later once your proof of concept is done.

[Lchen.11]

Yes, I know by STM32Cube I can filter down and get a possible hardware, but in my company I can’t install the tools temporary.

Could you just give a suggestion on which platform?

I need a DCMI(Camera), but I still confused which kind of Camera I can choose, because I want to realize the feature of the camera can record the speed of the handwriting.

[Lchen.11]

​thanks your help. Now I already use STM32Cube get one chip STM32F429XX with STM32429I-EVAL board choose. but in my project has one requirements is that the camera can record the speed of handwritten. so I suppose this work should can be done by X-CUBE_AI tool. Am I right? and in STM32Cube I can see six models: Keras,  TFLite,ONNX, Lasagne, Caffe, Convnetjs. I am do not know the difference among them. Could you let me know which one I should choose? or if somewhere I am not correct.