TensorFlow Read TFRecords Posted on 2019-10-20 | In BigData , Framework , Algorithm | Example Code: 123456789101112131415161718192021import tensorflow as tf record_path = '/tmp/tfrecord_write_test/part-r-00000' filename_queue = tf.train.string_input_producer([record_path]) reader = tf.TFRecordReader() _, serialized_example = reader.read(filename_queue) features = tf.parse_single_example(serialized_example, features={ 'label': tf.FixedLenFeature([], tf.float32), 'features': tf.FixedLenFeature([2], tf.float32), }) with tf.Session() as sess: init_op = tf.initialize_all_variables() sess.run(init_op) coord = tf.train.Coordinator() threads = tf.train.start_queue_runners(coord=coord) for i in range(2): example = sess.run([features])#在会话中取出image和label print(example) Example Data: https://alphablacktan.github.io/resource/post_attachment/tfrecord_write_test