The inference speed of the model is 3.6 seconds per image. Isn't that too slow? 
The inference speed of the model is 3.6 seconds per image. Isn't that too slow?