 Jérôme BUISINE
|
0c91d181de
update log information
|
3 anni fa |
|
Jérôme BUISINE
|
e4efb39660
fix pixel per pixel normalization
|
3 anni fa |
|
Jérôme BUISINE
|
edd3e59e37
normalization of pixel
|
3 anni fa |
|
Jérôme BUISINE
|
6a890c4fcb
try using 5 dim
|
3 anni fa |
|
Jérôme BUISINE
|
2e97e742c9
fix folder creation
|
3 anni fa |
|
Jérôme BUISINE
|
6caaae83ee
default resize
|
3 anni fa |
|
Jérôme BUISINE
|
5363b58ded
Load test dataset later
|
3 anni fa |
|
Jérôme BUISINE
|
8263f16a90
use of float16 instead
|
3 anni fa |
|
Jérôme BUISINE
|
124af56005
use of datasets path instead
|
3 anni fa |
|
Jérôme BUISINE
|
13e29e2a61
Add description information
|
3 anni fa |
|
Jérôme BUISINE
|
56b71d2491
update way of saving model using Keras API
|
4 anni fa |
|
Jérôme BUISINE
|
153f06bb97
model prediction updated for LSTM
|
4 anni fa |
|
Jérôme BUISINE
|
e3e10a84c8
Reduce data amount using float32 value
|
4 anni fa |
|
Jérôme BUISINE
|
1f55a401f3
specify chanels images input for window in LSTM
|
4 anni fa |
|
Jérôme BUISINE
|
24d5230ef5
update cnn parameters
|
4 anni fa |
|
Jérôme BUISINE
|
70f5d74832
- add models backups for LSTM
|
4 anni fa |
|
Jérôme BUISINE
|
ae38dc94c9
Split whole data into train, val and test
|
4 anni fa |
|
Jérôme BUISINE
|
7b022b2243
change optimizer parameter
|
4 anni fa |
|
Jérôme BUISINE
|
9f445fa2a5
update lstm model architecture
|
4 anni fa |
|
Jérôme BUISINE
|
4a441bc1c3
update number of epochs
|
4 anni fa |
|
Jérôme BUISINE
|
f076c19296
update dropout for model and saves of model results
|
4 anni fa |
|
Jérôme BUISINE
|
f3912184f8
remove dropout from Conv3D layer
|
4 anni fa |
|
Jérôme BUISINE
|
df4acbba4a
add of dropout for LSTM 2D layers
|
4 anni fa |
|
Jérôme BUISINE
|
7b6a3ed93b
input normalization
|
4 anni fa |
|
Jérôme BUISINE
|
a9fa043e7c
update model params
|
4 anni fa |
|
Jérôme BUISINE
|
a7545babe8
Prediction scripts added for CNN and LSTM
|
4 anni fa |
|
Jérôme BUISINE
|
e9609c0d47
add first version of LSTM 2D model
|
4 anni fa |
