search

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
2.23k stars 1.24k forks source link

TensorRT #475

Open tiagodavi opened 1 year ago

tiagodavi commented 1 year ago

First of all, thank you for the implementation.

Well, most of those script files have a line like bellow, but it's using trt nowhere flags.DEFINE_string('framework', 'tf', '(tf, tflite, trt')

Where is FLAGS.framework == 'trt'?

benchmarks.py has an implementation for TRT, but it still depends on the regular tensorflow/tf ? 

 elif FLAGS.framework == 'trt':
        saved_model_loaded = tf.saved_model.load(FLAGS.weights, tags=[tag_constants.SERVING])
        signature_keys = list(saved_model_loaded.signatures.keys())
        print(signature_keys)
        infer = saved_model_loaded.signatures['serving_default']

My question is, how to use tensorRT only without relying on tensorflow?

And if someone sees this issue and know how to install tensforflow 2 on jetson xavier + balenaOS with gpu please leave the link, because I have been trying to make it work and it's pain! For now I have trt installed, but not the normal tensorflow.