Closed BSN4 closed 5 years ago
You can use the TinyYOLOv3 model and by setting detector.setModelTypeAsTinyYOLOv3() and downloading the model file in the link below.
https://github.com/OlafenwaMoses/ImageAI/releases/download/1.0/yolo-tiny.h5
Please not that there is limit to the speed your non-GPU system can achieve. Also note that accuracy decreases as speed increases.
Thank you for your great effort I decided to use Azure API for now until my hosting provider release affordable cloud GPU server
using Linux and resnet without gpu make detection time very slow I'm waiting for 16 seconds only for just one image. can I make it more faster ?