Microcontroller + Camera solution | RPI alternatives

Hello,
I've created an app PC <->Teensy 4.0 <-> Arducam OV5642 5MP camera.
Transferring even a 640X480 image to the PC takes around 4 seconds, a 5MP image takes around 10-15 seconds.
The OV5642 shield SPI speed is 8MHZ which I assume is the bottleneck.

RPI + 5MP RPI cameras take high quality pictures within a much shorter timeframe.

1. Is there a microcontroller + embedded camera able to quickly capture and trasnfer an image back to the pc? (with/without the processing of the image happening on the controller)
2. Is there a microcontroller I can connect to the RPI 5MP CSI camera to achieve similar results to that of the RPI?

Note - my limitations are no SD/wifi/bluetooth etc.
Image has to be transferred through the USB connection.

I realize a microcontroller is not a single board computer and RPI has a dedicated gpu, yet I am wondering what can be achieved on a microcontroller level.

Thanks in advance!