Description
I wish some modules for more efficiently data reading : like tf.data / tf.train.slice_input_producer / tf.train.string_input_producer . If possible . Thank you very much!
Issue background : I want to load local images for train , in order to improve the efficiency of CPU and GPU,I hope to use multithreading and tf pipeline .
Use case 1#: tf.data
Describe: I hear that google suggest to use tf.data for more efficiently data reading , and in tf 2.0 , tf.data is officially designated . So I wish tf .NET can add tf.date in the future . thank you~
Url of tf.data document :
https://tensorflow.google.cn/guide/data?hl=zh-cn
https://tensorflow.google.cn/tutorials/load_data/images?hl=zh-cn
https://github.com/tensorflow/docs/blob/master/site/zh-cn/tutorials/load_data/images.ipynb
Use case 2#: tf.train.slice_input_producer
tf.train.string_input_producer
Describe: we use tf.train.slice_input_producer or tf.train.string_input_producer to create image files queue ,then use tf.train.start_queue_runners to fill queue , and use Coordinator to manage thread . I wish tf .NET can add these in the future . thank you~
Url of these document :
https://tensorflow.google.cn/versions/r1.15/api_docs/python/tf/train/slice_input_producer
https://tensorflow.google.cn/versions/r1.15/api_docs/python/tf/train/string_input_producer