Updated the saving log and model functions, added the experiment of deepEye,...

Updated the saving log and model functions, added the experiment of deepEye, and make get_data function in IOHelper more generalised

CNN/CNN.py

@@ -45,9 +45,9 @@ def run(trainX, trainY):
     # PATH = '../cifar_net.pth'
     # torch.save(net.state_dict(), PATH)
     # Newly added lines below
-    save_logs(hist, pytorch=True)
-    save_model_param(pytorch=True)
-    plot_loss_torch(loss)
+	save_logs(hist, config['model_dir'], config['model'], pytorch=True)
+	save_model_param(config['model_dir'], config['model'], pytorch=True)
+	plot_loss_torch(loss)
 
 def train(trainloader, net, optimizer, criterion, epoch=50):
     loss=[]

DeepEye/deepEye.py

@@ -12,14 +12,14 @@ import seaborn as sns
 sns.set_style('darkgrid')
 
 
-def run(trainX,trainY):
+def run(trainX, trainY):
     classifier = Classifier_DEEPEYE(output_directory=config['root_dir'], input_shape=(129, 500))
     hist = classifier.fit(deepeye_x=trainX, y=trainY)
     plot_loss(hist, 'DeepEye', True)
     plot_acc(hist, 'DeepEye', True)
     # Newly added lines below
-    save_logs(hist, pytorch=False)
-    save_model_param(pytorch=False)
+    save_logs(hist, config['model_dir'], config['model'], pytorch=False)
+    save_model_param(config['model_dir'], config['model'], pytorch=False)
 
 class Classifier_DEEPEYE:
     """
@@ -40,7 +40,7 @@ class Classifier_DEEPEYE:
       y_pred          : class (left/right for nb_class=2) of the given input_tensor
     """
 
-    def __init__(self, output_directory, input_shape, nb_classes, verbose=False, build=True,
+    def __init__(self, output_directory, input_shape, verbose=False, build=True,
                  batch_size=64, nb_filters=32, use_residual=True, use_bottleneck=True, depth=6,
                  kernel_size=41, nb_epochs=1500):
 

EEGNet/eegNet.py

@@ -18,8 +18,8 @@ def run(trainX, trainY):
     plot_loss(hist, 'EEGNet', True)
     plot_acc(hist, 'EEGNet', True)
     # Newly added lines below
-    save_logs(hist, pytorch=False)
-    save_model_param(pytorch=False)
+	save_logs(hist, config['model_dir'], config['model'], pytorch=False)
+	save_model_param(config['model_dir'], config['model'], pytorch=False)
 
 class Classifier_EEGNet:
     def __init__(self, output_directory, nb_classes=1, chans = 129, samples = 500, dropoutRate = 0.5, kernLength = 64, F1 = 8,

InceptionTime/inception.py

@@ -11,8 +11,8 @@ def run(trainX, trainY):
     plot_loss(hist, 'Inception', True)
     plot_acc(hist, 'Inception', True)
     # Newly added lines below
-    save_logs(hist, pytorch=False)
-    save_model_param(pytorch=False)
+	save_logs(hist, config['model_dir'], config['model'], pytorch=False)
+	save_model_param(config['model_dir'], config['model'], pytorch=False)
 
 class Classifier_INCEPTION:
     def __init__(self, output_directory, input_shape, verbose=False, build=True, batch_size=64, nb_filters=32,

README.md

 # DL-Project
 
 Predicting eye gaze with DL
-    Projects that we try: CNN, InceptionTime, EEGNet, DeepEye
\ No newline at end of file
+    Projects that we try: CNN, InceptionTime, EEGNet, DeepEye
+
+
+
+## Leonhard Settings
+
+Please use the command
+
+`module load gcc/6.3.0 python_gpu/3.8.5`
+
+before training the model. Please make sure that the Tensorflow version should be 2.x.
\ No newline at end of file

config.py

@@ -15,7 +15,7 @@ config['log_dir'] = './runs/'
 config['tmp_dir'] = './tmp/'
 # Path to training, validation and test data folders.
 # PS: Note that we have to upload the data to the server!!!
-config['data_dir'] = '/Volumes/methlab/ETH_AS/preprocessed2/'
+config['data_dir'] = '/cluster/home/your_username/data/'
 config['data_dir_server'] = '/cluster/project/infk/zigeng/preprocessed2/'
 # Path of root
 config['root_dir'] = '.'

main.py

@@ -10,7 +10,7 @@ import numpy as np
 def main():
     start_time = time.time()
     try:
-        trainX, trainY = IOHelper.get_data(verbose=True)
+        trainX, trainY = IOHelper.get_data(config['data_dir'], verbose=True)
         # IOHelper.store(trainX, trainY)
     except:
         return

utils/IOHelper.py

@@ -4,15 +4,15 @@ import scipy.io as sio
 import os
 import pickle
 
-def get_data(verbose=True):
-    pkl_file_x = open('x.pkl', 'rb')
+def get_data(data_dir, verbose=True):
+    pkl_file_x = open(data_dir + 'x.pkl', 'rb')
     x = pickle.load(pkl_file_x)
     pkl_file_x.close()
     if verbose:
         print("X training loaded.")
         print(x.shape)
 
-    pkl_file_y = open('y.pkl', 'rb')
+    pkl_file_y = open(data_dir + 'y.pkl', 'rb')
     y = pickle.load(pkl_file_y)
     pkl_file_y.close()
     if verbose:

utils/utils.py

-from config import config
 import matplotlib.pyplot as plt
 import seaborn as sns
 import numpy as np
 import torch
 import pandas as pd
+import os
 sns.set_style('darkgrid')
 
 def plot_acc(hist,name,val=False):
@@ -50,22 +50,21 @@ def plot_loss_torch(loss,name='CNN'):
     plt.show()
 
 
-# Save the logs (newly added without debugging)
-# haven't write about the pytorch=True part 
-def save_logs(hist, output_directory=config['model_dir'], pytorch=False):
+# Save the logs  
+def save_logs(hist, output_directory, model, pytorch=False):
     if pytorch:
-        df_metrics={'Loss':hist}
-        df_metrics=pd.DataFrame(df_metrics)
-        df_metrics.to_csv(output_directory + config['model'] + '_df_metrics.csv', index=False)
+        df_metrics = {'Loss':hist}
+        df_metrics = pd.DataFrame(df_metrics)
+        df_metrics.to_csv(output_directory + '/' + model + '_' + 'df_metrics.csv', index=False)
         
 
     else:
         hist_df = pd.DataFrame(hist.history)
-        hist_df.to_csv(output_directory + config['model'] + '_history.csv', index=False)
+        hist_df.to_csv(output_directory + '/' + model + '_' + 'history.csv', index=False)
         
         df_metrics = {'Accuracy': hist_df['accuracy'], 'Loss': hist_df['loss']}
         df_metrics = pd.DataFrame(df_metrics)
-        df_metrics.to_csv(output_directory + config['model'] + '_df_metrics.csv', index=False)
+        df_metrics.to_csv(output_directory + '/' + model + '_' + 'df_metrics.csv', index=False)
 
         index_best_model = hist_df['loss'].idxmin()
         row_best_model = hist_df.loc[index_best_model]
@@ -77,13 +76,12 @@ def save_logs(hist, output_directory=config['model_dir'], pytorch=False):
         df_best_model['best_model_val_loss'] = row_best_model['val_loss']
         df_best_model['best_model_train_acc'] = row_best_model['acc']
         df_best_model['best_model_val_acc'] = row_best_model['val_acc']
-        
-        df_best_model.to_csv(output_directory + config['model'] + '_df_best_model.csv', index=False)
+        df_best_model.to_csv(output_directory + '/' + model + '_' + 'df_best_model.csv', index=False)
 
 # Save the model parameters (newly added without debugging)
-def save_model_param(output_directory=config['model_dir'], pytorch=False):
+def save_model_param(output_directory, model, pytorch=False):
     if pytorch:
-        torch.save(net.state_dict(), output_directory + config['model'] + '_model.pth')
+        torch.save(net.state_dict(), output_directory + '/' + model + '_' + 'model.pth')
     else:
-        classifier.save(output_directory + config['model'] + '_model.h5')
+        classifier.save(output_directory + '/' + model + '_' + 'model.h5')