Notice
Recent Posts
Recent Comments
Link
일 | 월 | 화 | 수 | 목 | 금 | 토 |
---|---|---|---|---|---|---|
1 | 2 | 3 | 4 | 5 | ||
6 | 7 | 8 | 9 | 10 | 11 | 12 |
13 | 14 | 15 | 16 | 17 | 18 | 19 |
20 | 21 | 22 | 23 | 24 | 25 | 26 |
27 | 28 | 29 | 30 |
Tags
- VGGnet
- Homomorphic Filter
- Embedded System
- OpenCV
- tf.train.match_filenames_once()
- IOT
- I.MX6Q
- Raspberry Pi
- cross compile
- TensorFlow
- deep-learning
- porting
- Python
- Machine learning
- Facial expression recognition
- Data Load
- tf.train.string_input_producer()
- CNN
- preprocessing
- Machine Vision
- tf.saver()
- ARM Processor
Archives
- Today
- Total
목록tf.train.string_input_producer() (1)
Austin's_Lab
[Tensorflow] Training data를 load하는 방법과 model을 save하다 발생한 문제
-Tensorflow 0.12.0 사용- Tensorflow에서 training image data를 load하기 위해서는 tf.train.string_input_producer()를 이용해 queue에 넣어줘야 한다. 외에도 TFRecord를 쓴다던지, binary file을 만들어 쓰거나 csv file format을 쓰는 등 다양한 방법이 있지만, 처음 Tensorflow를 접했을 땐 tf.train.string_input_producer를 찾은 것만으로도 굉장한 수확이었다. Data가 많아질수록 TFRecord를 사용하는게 훨씬 빠르다고는 하던데 학습 모델을 만들고 hyper parameter 조절이나 데이터 전처리 작업 연구에 집중하다보니 TFRecord를 만들고 사용하는 법에 대해 찾아볼 시간..
Machine Learning
2017. 3. 16. 22:09