high inference time running yolov4-416-fp32.tflite model with Android example code

hunglc007 / tensorflow-yolov4-tflite

YOLOv4, YOLOv4-tiny, YOLOv3, YOLOv3-tiny Implemented in Tensorflow 2.0, Android. Convert YOLO v4 .weights tensorflow, tensorrt and tflite

https://github.com/hunglc007/tensorflow-yolov4-tflite

MIT License

2.24k stars 1.24k forks source link

high inference time running yolov4-416-fp32.tflite model with Android example code #299

Open andophine opened 3 years ago

andophine commented 3 years ago

Using Android sample code running on Qualcomm SDM660 CPU, high inference time is obtained, it take more than 300 msec for INPUT_SIZE 416 per frame. Code I have changed is setting the flag isTiny as true in YoloV4Classifier.java to allow code running normally.

Compare result with google TFLite Object Detection Demo App that consume less than 60 ms per frame with INPUT_SIZE 300. Any one know this issue?

Thanks Angus