From 09f9b55ea8a49f7ea5f4aaa685da72d09556085a Mon Sep 17 00:00:00 2001
From: "Daniel.Frisinghelli" <daniel.frisinghelli@eurac.edu>
Date: Wed, 13 Jan 2021 18:04:23 +0100
Subject: [PATCH] Implemented changes of the configurable parameters.
---
pysegcnn/main/config.py | 42 +++++++++++++++++++++--------------------
1 file changed, 22 insertions(+), 20 deletions(-)
diff --git a/pysegcnn/main/config.py b/pysegcnn/main/config.py
index 939938d..015f663 100644
--- a/pysegcnn/main/config.py
+++ b/pysegcnn/main/config.py
@@ -50,8 +50,8 @@ TILE_SIZE = None
# the source dataset configuration dictionary
src_ds_config = {
- # ------------------------------- Dataset ---------------------------------
-
+ # -------------------------------------------------------------------------
+ # Dataset -----------------------------------------------------------------
# -------------------------------------------------------------------------
# name of the dataset
@@ -166,6 +166,10 @@ trg_ds_config = {
# the source dataset split configuration dictionary
src_split_config = {
+ # -------------------------------------------------------------------------
+ # Dataset split -----------------------------------------------------------
+ # -------------------------------------------------------------------------
+
# the mode to split the dataset:
#
# - 'random': randomly split the scenes
@@ -200,7 +204,7 @@ src_split_config = {
# scenes before date build the training set, scenes after the date build
# the validation set, the test set is empty
# used if split_mode='date'
- 'date': '20161231',
+ 'date': '',
'dateformat': '%Y%m%d',
# whether to drop samples (during training only) with a fraction of
@@ -217,12 +221,12 @@ src_split_config = {
# the target dataset split configuration dictionary
trg_split_config = {
- 'split_mode': 'date',
+ # 'split_mode': 'date',
# 'split_mode': 'random',
- # 'split_mode': 'scene',
+ 'split_mode': 'scene',
'ttratio': 1,
'tvratio': 0.8,
- 'date': '20171231',
+ 'date': '',
'dateformat': '%Y%m%d',
'drop': 0,
}
@@ -230,14 +234,16 @@ trg_split_config = {
# the model configuration dictionary
model_config = {
- # ------------------------------ Network ----------------------------------
-
+ # -------------------------------------------------------------------------
+ # Network -----------------------------------------------------------------
# -------------------------------------------------------------------------
# define the model
'model_name': 'Segnet',
+ # -------------------------------------------------------------------------
# Transfer learning -------------------------------------------------------
+ # -------------------------------------------------------------------------
# whether to apply any sort of transfer learning
# if transfer=False, the model is only trained on the source dataset
@@ -247,8 +253,7 @@ model_config = {
# name of the pretrained model to apply for transfer learning
# Required if supervised=True
# Optional if unsupervised=True
- 'pretrained_model': 'Unet_SparcsDataset_Adam_RandomSplit_s0_t10v08_t125_b128_r4g3b2n5.pt', # nopep8
- # 'pretrained_model': 'Unet_SparcsDataset_Adam_SceneSplit_s0_t10v08_t125_b128_r4g3b2n5.pt', # nopep8
+ 'pretrained_model': '', # nopep8
# Supervised vs. Unsupervised ---------------------------------------------
@@ -277,13 +282,15 @@ model_config = {
# the layer where to compute the domain adaptation loss
# currently, the following positions are supported:
+ # - 'inp': compute the domain adaptation loss with the input features
# - 'enc': compute the domain adaptation loss with the encoder features
# - 'dec': compute the domain adaptation loss with the decoder features
# - 'cla': compute the domain adaptation loss with the classifier
# features
- # 'uda_pos': 'enc',
+ # 'uda_pos': 'inp',
+ 'uda_pos': 'enc',
# 'uda_pos': 'dec',
- 'uda_pos': 'cla',
+ # 'uda_pos': 'cla',
# The weight of the domain adaptation, trading off adaptation with
# classification accuracy on the source domain.
@@ -299,8 +306,8 @@ model_config = {
# the source domain.
'uda_lambda': 0.01,
- # ----------------------------- Training ---------------------------------
-
+ # -------------------------------------------------------------------------
+ # Training configuration --------------------------------------------------
# -------------------------------------------------------------------------
# whether to save the model state to disk
@@ -330,9 +337,6 @@ model_config = {
# the whole training dataset
'epochs': 100,
- # define a classification loss function to calculate the network error
- 'cla_loss': 'CrossEntropy',
-
# define an optimizer to update the network weights
'optim_name': 'Adam',
@@ -342,14 +346,12 @@ model_config = {
# optimizer keyword arguments
'optim_kwargs': {'weight_decay': 0.01, # the weight decay rate
'amsgrad': False} # whether to use AMSGrad variant
-
}
# the evaluation configuration file
eval_config = {
-
+ # -------------------------------------------------------------------------
# ----------------------------- Evaluation --------------------------------
-
# -------------------------------------------------------------------------
# these options are only used for evaluating a trained model using
--
GitLab