I was addicted to an error when trying to process asynchronously with tensorflow. I searched for a workaround, but couldn't find it, so I hope it helps someone.
Running the following code causes an error in tf.train.start_queue_runners ()
filename_queue = tf.train.string_input_producer(["../dataset/sample1.tfrecord"])
reader = tf.TextLineReader()
// ...Abbreviation
with tf.Session() as sess:
#Input enqueue thread start
coord = tf.train.Coordinator()
threads = tf.train.start_queue_runners(sess=sess, coord=coord)
[Execution result]
Process finished with exit code -1073741819 (0xC0000005)
The reference destination of the file path was incorrect.
(Wrong)filename_queue = tf.train.string_input_producer(["../dataset/sample1.tfrecord"])
(Positive)filename_queue = tf.train.string_input_producer(["../dataset/samples/sample1.tfrecord"])
It's just a mistake, but I'm addicted to it.
I wish I could get an error with tf.train.string_input_producer ()
...
Since 0xC0000005
is a memory access violation type error in the first place, you should first doubt the reference surroundings such as files.
The following article is easy to understand. https://qiita.com/antimon2/items/c7d2285d34728557e81d https://qiita.com/ashigirl966/items/99b0f8d9713ee90db13a