in_memory: false does not work with images
See original GitHub issueDescribe the bug
the preprocesing setting in_memory: false
for images as input features does not seem to work. The hdf5 file is still loaded into the RAM during training.
To Reproduce a .yaml file like this one:
input_features:
-
name: image_path
type: image
preprocessing:
in_memory: false
encoder: stacked_cnn
conv_layers:
-
num_filters: 32
filter_size: 15
pool_size: 2
pool_stride: 2
-
num_filters: 64
filter_size: 15
pool_size: 2
pool_stride: 2
dropout: true
fc_layers:
-
fc_size: 128
dropout: true
output_features:
-
name: class
type: category
training:
reduce_learning_rate_on_plateau: 10
early_stop: 10
learning_rate: 0.0004
batch_size: 40
Same is true when I use preprocessing as a separate section at the end of the .yaml file.
Issue Analytics
- State:
- Created 4 years ago
- Comments:12
Top Results From Across the Web
bug: Unable to set -inMemory for Dynamodb #6506 - GitHub
Use case: I'm running tests on an ephemeral localstack container and don't need persistence. I think I'm experiencing the performance hit in # ......
Read more >How to display image during in-memory AngularJS processing
Even when I set loading to false inside the loadTable function, the image never appears. When I comment out loading = false, the...
Read more >Configuring the Apollo Client cache - Apollo GraphQL Docs
Disabling normalization. You can instruct the InMemoryCache not to normalize objects of a particular type. This can be useful for metrics and other...
Read more >Why? Trying to write jpeg in memory and save to file does ...
I'm trying to do the following: 1. Create a memory buffer with valid jpg image. 2. Add Exif data to the jpg image...
Read more >Datastore for image data - MATLAB - MathWorks
Use an ImageDatastore object to manage a collection of image files, where each individual image fits in memory, but the entire collection of...
Read more >Top Related Medium Post
No results found
Top Related StackOverflow Question
No results found
Troubleshoot Live Code
Lightrun enables developers to add logs, metrics and snapshots to live code - no restarts or redeploys required.
Start FreeTop Related Reddit Thread
No results found
Top Related Hackernoon Post
No results found
Top Related Tweet
No results found
Top Related Dev.to Post
No results found
Top Related Hashnode Post
No results found
Top GitHub Comments
A related project. I am trying to identify individuals from the same worm species based on their pigmentation pattern. Face recognition for worms so to speak.
@tboo can you provide more details please? Otherwise it would be difficult for us to reproduce and solve. @ydudin3 can you look into this?