reformat files

YaelBenShalom · Oct 26, 2021 · 01c8dc1 · 01c8dc1
1 parent b859b35
commit 01c8dc1
Show file tree

Hide file tree

Showing 18 changed files with 447 additions and 337 deletions.
diff --git a/README.md b/README.md
@@ -1,35 +1,35 @@
 # Objects Recognition and Classification
+
 ## ME499 - Independent Project, Winter 2021
+
 Yael Ben Shalom, Northwestern University.
 
+## Table of Contents
 
-Table of Contents
------------------
-  * [Project Overview](#project-overview)
-    * [Traffic sign detection and classification](#traffic-sign-detection-and-classification)
-    * [Trash detection, classification, and segmentation](#trash-detection-classification-and-segmentation)
+- [Project Overview](#project-overview)
+  - [Traffic sign detection and classification](#traffic-sign-detection-and-classification)
+  - [Trash detection, classification, and segmentation](#trash-detection-classification-and-segmentation)
 
 ## Project Overview
-This project contains 2 modules:
 
+This project contains 2 modules:
 
 ### [**Traffic sign detection and classification**](https://github.com/YaelBenShalom/Objects-Recognition-and-Classification/tree/master/traffic_signs_detection)
 
 In this module I built and trained a neural network to detect and classify different traffic signs using PyTorch, OpenCV and YoloV3.<br>
 This module contained 2 parts:
-    * [traffic sign classification](https://github.com/YaelBenShalom/Objects-Recognition-and-Classification/tree/master/traffic_signs_detection/traffic_signs_classification) - Were I build and trained a neural network to classify different traffic-signs images.
-    * [traffic sign detection](https://github.com/YaelBenShalom/Objects-Recognition-and-Classification/tree/master/traffic_signs_detection/traffic_signs_recognition) - Were I trained neural network to detect different traffic-signs in an image/video.
+_ [traffic sign classification](https://github.com/YaelBenShalom/Objects-Recognition-and-Classification/tree/master/traffic_signs_detection/traffic_signs_classification) - Were I build and trained a neural network to classify different traffic-signs images.
+_ [traffic sign detection](https://github.com/YaelBenShalom/Objects-Recognition-and-Classification/tree/master/traffic_signs_detection/traffic_signs_recognition) - Were I trained neural network to detect different traffic-signs in an image/video.
 
     An example of traffic detection program output:<br>
     ![Traffic-Sign Recognition](https://github.com/YaelBenShalom/Objects-Recognition-and-Classification/blob/master/traffic_signs_detection/traffic_signs_recognition/images/traffic-sign.gif)
 
-
 ### [**Trash detection, classification, and segmentation**](https://github.com/YaelBenShalom/Objects-Recognition-and-Classification/tree/master/trash_detection)
 
 In this module I built and trained a neural network to detect and classify different traffic signs using PyTorch, OpenCV and YoloV5.<br>
 This module contained 2 parts:
-    * [trash classification](https://github.com/YaelBenShalom/Objects-Recognition-and-Classification/tree/master/traffic_signs_detection/traffic_signs_classification) - Were I build and trained a neural network to classify different recyclable objects' images.
-    * [trash detection](https://github.com/YaelBenShalom/Objects-Recognition-and-Classification/tree/master/traffic_signs_detection/traffic_signs_recognition) - Were I trained neural network to detect different recyclable objects in an image/video.
+_ [trash classification](https://github.com/YaelBenShalom/Objects-Recognition-and-Classification/tree/master/traffic_signs_detection/traffic_signs_classification) - Were I build and trained a neural network to classify different recyclable objects' images.
+_ [trash detection](https://github.com/YaelBenShalom/Objects-Recognition-and-Classification/tree/master/traffic_signs_detection/traffic_signs_recognition) - Were I trained neural network to detect different recyclable objects in an image/video.
 
     An example of trash detection program output:<br>
     ![Trash Recognition](https://github.com/YaelBenShalom/Objects-Recognition-and-Classification/blob/master/trash_detection/trash_recognition/images/real-time%20detection2.gif)

diff --git a/traffic_signs_detection/traffic_signs_classification/README.md b/traffic_signs_detection/traffic_signs_classification/README.md
@@ -1,80 +1,82 @@
 # Traffic-Sign Classification
+
 ## ME499 - Independent Project, Winter 2021
+
 Yael Ben Shalom, Northwestern University.<br>
 This module is a part of a [Objects Recognition and Classification](https://github.com/YaelBenShalom/Objects-Recognition-and-Classification) project.
 
+## Table of Contents
 
-Table of Contents
------------------
-  * [Module Overview](#module-overview)
-  * [User Guide](#user-guide)
-    * [Program Installation](#program-installation)
-    * [Quickstart Guide](#quickstart-guide)
-  * [Dataset](#dataset)
-
+- [Module Overview](#module-overview)
+- [User Guide](#user-guide)
+  - [Program Installation](#program-installation)
+  - [Quickstart Guide](#quickstart-guide)
+- [Dataset](#dataset)
 
 ## Module Overview
+
 In this module I built and trained a neural network to classify different traffic signs using PyTorch.<br>
 I based my program on the German Traffic Sign Recognition Benchmark ([GTSRB](https://benchmark.ini.rub.de/gtsrb_news.html)) dataset.
 
-
 ## User Guide
+
 ### Program Installation
 
 1. Clone the repository, using the following commands:
-    ```
-    git clone https://github.com/YaelBenShalom/Objects-Recognition-and-Classification/tree/master/traffic_signs_detection/traffic_signs_classification
-    ```
 
-2. Download the dataset and extract it into `./data` directory. The dataset can be found on the [INI Benchmark website](https://benchmark.ini.rub.de/?section=gtsrb&subsection=news), or downloaded directly through [here](https://s3-us-west-1.amazonaws.com/udacity-selfdrivingcar/traffic-signs-data.zip).
+   ```
+   git clone https://github.com/YaelBenShalom/Objects-Recognition-and-Classification/tree/master/traffic_signs_detection/traffic_signs_classification
+   ```
 
+2. Download the dataset and extract it into `./data` directory. The dataset can be found on the [INI Benchmark website](https://benchmark.ini.rub.de/?section=gtsrb&subsection=news), or downloaded directly through [here](https://s3-us-west-1.amazonaws.com/udacity-selfdrivingcar/traffic-signs-data.zip).
 
 ### Quickstart Guide
 
 Run the classification program:
+
 1. To train and test the program on the dataset, run the following command from the root directory:
-    ```
-    python code/classification.py
-    ```
+
+   ```
+   python code/classification.py
+   ```
 
 2. To train the program on the dataset and test it on a specific image, copy the image to the root directory and run the following command from the root directory:
-    ```
-    python code/classification.py --image <image-name>
-    ```
-    Where `<image-name>` is the name of the image (including image type).<br>
-    The trained model will be saved in the root directory as `/model`.
 
-3. To to use an existing model and test it on a specific image, copy the image to the root directory and run the following command from the root directory:
-    ```
-    python code/classification.py --image <image-name> --model <model-name>
-    ```
-    Where `<model-name>` is the name of the trained model.
+   ```
+   python code/classification.py --image <image-name>
+   ```
 
+   Where `<image-name>` is the name of the image (including image type).<br>
+   The trained model will be saved in the root directory as `/model`.
+
+3. To to use an existing model and test it on a specific image, copy the image to the root directory and run the following command from the root directory:
+   ```
+   python code/classification.py --image <image-name> --model <model-name>
+   ```
+   Where `<model-name>` is the name of the trained model.
 
 <br>The program output when running it on the example image:
 
 The loss plot:<br>
-    ![Loss Graph](https://github.com/YaelBenShalom/Objects-Recognition-and-Classification/blob/master/traffic_signs_detection/traffic_signs_classification/images/Losses%20(100%20Epochs).png)
-
+![Loss Graph](<https://github.com/YaelBenShalom/Objects-Recognition-and-Classification/blob/master/traffic_signs_detection/traffic_signs_classification/images/Losses%20(100%20Epochs).png>)
 
 The accuracy plot:<br>
-    ![Accuracy Graph](https://github.com/YaelBenShalom/Objects-Recognition-and-Classification/blob/master/traffic_signs_detection/traffic_signs_classification/images/Accuracy%20(100%20Epochs).png)
-
+![Accuracy Graph](<https://github.com/YaelBenShalom/Objects-Recognition-and-Classification/blob/master/traffic_signs_detection/traffic_signs_classification/images/Accuracy%20(100%20Epochs).png>)
 
 The output image (with the correct prediction):<br>
-    ![Accuracy Graph](https://github.com/YaelBenShalom/Objects-Recognition-and-Classification/blob/master/traffic_signs_detection/traffic_signs_classification/images/Image_Classification.png)
-
+![Accuracy Graph](https://github.com/YaelBenShalom/Objects-Recognition-and-Classification/blob/master/traffic_signs_detection/traffic_signs_classification/images/Image_Classification.png)
 
 ## Dataset
 
 The German Traffic Sign Recognition Benchmark ([GTSRB](https://benchmark.ini.rub.de/gtsrb_news.html)) is a large multi-category classification benchmark. It was used in a competition at the International Joint Conference on Neural Networks (IJCNN) 2011.
 
 The benchmark has the following properties:
+
 1. Single-image, multi-class classification problem.
 2. 43 classes.
 3. More than 50,000 images in total (~35,000 training images, ~4000 validation images, and ~13,000 testing images).
 4. Large, lifelike database.
 
 The dataset can be found on the [INI Benchmark website](https://benchmark.ini.rub.de/?section=gtsrb&subsection=news), or downloaded directly through [here](https://s3-us-west-1.amazonaws.com/udacity-selfdrivingcar/traffic-signs-data.zip).
 
-![The German Traffic Sign Recognition Benchmark](https://github.com/YaelBenShalom/Traffic-Sign-Recognition-and-Classification/blob/master/traffic_signs_detection/traffic_signs_classification/images/dataset.png)
+![The German Traffic Sign Recognition Benchmark](https://github.com/YaelBenShalom/Traffic-Sign-Recognition-and-Classification/blob/master/traffic_signs_detection/traffic_signs_classification/images/dataset.png)
diff --git a/traffic_signs_detection/traffic_signs_classification/code/classification.py b/traffic_signs_detection/traffic_signs_classification/code/classification.py
@@ -47,9 +47,12 @@ def dataset_properties(trainset_name, validset_name, testset_name, class_names,
 
     Output: None
     """
-    train_features, train_labels = load_dataset(trainset_name, base_folder=data_dir)
-    valid_features, valid_labels = load_dataset(validset_name, base_folder=data_dir)
-    test_features, test_labels = load_dataset(testset_name, base_folder=data_dir)
+    train_features, train_labels = load_dataset(
+        trainset_name, base_folder=data_dir)
+    valid_features, valid_labels = load_dataset(
+        validset_name, base_folder=data_dir)
+    test_features, test_labels = load_dataset(
+        testset_name, base_folder=data_dir)
 
     # print(f"train_features shape: {train_features.shape}")
     # print(f"train_labels shape: {train_labels.shape}")
@@ -85,7 +88,8 @@ def dataset_properties(trainset_name, validset_name, testset_name, class_names,
     for i in range(40):
         feature_index = random.randint(0, train_labels.shape[0])
         plt.subplot(6, 8, i + 1)
-        plt.subplots_adjust(left=0.1, bottom=0.03, right=0.9, top=0.92, wspace=0.2, hspace=0.2)
+        plt.subplots_adjust(left=0.1, bottom=0.03, right=0.9,
+                            top=0.92, wspace=0.2, hspace=0.2)
         plt.axis('off')
         plt.imshow(train_features[feature_index])
     plt.suptitle('Random Training Images', fontsize=20)
@@ -97,7 +101,8 @@ def dataset_properties(trainset_name, validset_name, testset_name, class_names,
     for i in range(classes_num):
         feature_index = random.choice(np.where(train_labels == i)[0])
         plt.subplot(6, 8, i + 1)
-        plt.subplots_adjust(left=0.1, bottom=0.03, right=0.9, top=0.92, wspace=0.2, hspace=0.2)
+        plt.subplots_adjust(left=0.1, bottom=0.03, right=0.9,
+                            top=0.92, wspace=0.2, hspace=0.2)
         plt.axis('off')
         plt.title(class_names[i], fontsize=10)
         plt.imshow(train_features[feature_index])
@@ -124,7 +129,7 @@ def class_names_fun(data_dir):
     for row in class_names_rows:
         label, label_name = row.strip().split(",")
         class_names[int(label)] = label_name
-    
+
     return class_names
 
 
@@ -136,14 +141,17 @@ def plot_training_results(train_loss_list, valid_loss_list, valid_accuracy_list,
       train_loss_list:      list of loss value on the entire training dataset.
       valid_loss_list:      list of loss value on the entire validation dataset.
       valid_accuracy_list:  list of accuracy on the entire validation dataset.
-    
+
     Output: None
     """
     # Plotting training and validation loss vs. epoch number
     plt.figure()
-    plt.plot(range(len(train_loss_list)), train_loss_list, label='Training Loss')
-    plt.plot(range(len(valid_loss_list)), valid_loss_list, label='Validation Loss')
-    plt.title(f'Training and Validation Loss Vs. Epoch Number ({epoch_num} Epochs)')
+    plt.plot(range(len(train_loss_list)),
+             train_loss_list, label='Training Loss')
+    plt.plot(range(len(valid_loss_list)),
+             valid_loss_list, label='Validation Loss')
+    plt.title(
+        f'Training and Validation Loss Vs. Epoch Number ({epoch_num} Epochs)')
     plt.xlabel('Epoch Number')
     plt.ylabel('Loss')
     plt.legend(loc="best")
@@ -152,7 +160,8 @@ def plot_training_results(train_loss_list, valid_loss_list, valid_accuracy_list,
 
     # Plotting validation accuracy vs. epoch number
     plt.figure()
-    plt.plot(range(len(valid_accuracy_list)), valid_accuracy_list, label='Validation Accuracy')
+    plt.plot(range(len(valid_accuracy_list)),
+             valid_accuracy_list, label='Validation Accuracy')
     plt.title(f'Validation Accuracy Vs. Epoch Number ({epoch_num} Epochs)')
     plt.xlabel('Epoch Number')
     plt.ylabel('Accuracy')
@@ -172,7 +181,7 @@ def main(args):
     """
 
     # Define dataset directory
-    data_dir  = "data"
+    data_dir = "data"
 
     # Define dataset files
     trainset_name = "train.p"
@@ -183,7 +192,8 @@ def main(args):
     class_names = class_names_fun(data_dir)
 
     # Visualizing the dataset
-    dataset_properties(trainset_name, validset_name, testset_name, class_names, data_dir)
+    dataset_properties(trainset_name, validset_name,
+                       testset_name, class_names, data_dir)
 
     # Define the device parameters
     torch.manual_seed(1)
@@ -201,7 +211,8 @@ def main(args):
     stop_threshold = 1e-4
 
     # Computing data transformation to normalize data
-    mean = (0.485, 0.456, 0.406)    # from https://pytorch.org/docs/stable/torchvision/transforms.html
+    # from https://pytorch.org/docs/stable/torchvision/transforms.html
+    mean = (0.485, 0.456, 0.406)
     std = (0.229, 0.224, 0.225)     # -"-
     transform = transforms.Compose([transforms.ToTensor(),
                                     transforms.Resize((32, 32)),
@@ -219,15 +230,16 @@ def main(args):
 
         # Train the network
         model, train_loss_list, valid_loss_list, valid_accuracy_list = run_model(model, running_mode='train',
-                                                                                train_set=train_dataset,
-                                                                                valid_set=valid_dataset,
-                                                                                test_set=test_dataset,
-                                                                                batch_size=batch_size, epoch_num=epoch_num,
-                                                                                learning_rate=learning_rate,
-                                                                                stop_thr=stop_threshold,
-                                                                                criterion=criterion, device=device)
+                                                                                 train_set=train_dataset,
+                                                                                 valid_set=valid_dataset,
+                                                                                 test_set=test_dataset,
+                                                                                 batch_size=batch_size, epoch_num=epoch_num,
+                                                                                 learning_rate=learning_rate,
+                                                                                 stop_thr=stop_threshold,
+                                                                                 criterion=criterion, device=device)
         # Plot the results of training the network
-        plot_training_results(train_loss_list, valid_loss_list, valid_accuracy_list, epoch_num)
+        plot_training_results(train_loss_list, valid_loss_list,
+                              valid_accuracy_list, epoch_num)
 
         # Save the trained model
         torch.save(model.state_dict(), model_path)
@@ -236,7 +248,7 @@ def main(args):
     else:
         # Defining the model
         model_path = os.path.abspath(args["model"])
-    
+
         # Load the trained model
         model.load_state_dict(torch.load(model_path, map_location=device))
 
@@ -263,7 +275,8 @@ def main(args):
 
         # Predict the class of the tested image
         prediction = int(predict(model, test_image_transform4d)[0])
-        print(f"Test prediction: {prediction} -> Class: {class_names[prediction]}")
+        print(
+            f"Test prediction: {prediction} -> Class: {class_names[prediction]}")
 
         # Plot the image with the predicted class
         plt.figure()

diff --git a/traffic_signs_detection/traffic_signs_classification/code/cnn.py b/traffic_signs_detection/traffic_signs_classification/code/cnn.py
@@ -50,8 +50,8 @@ def __init__(self):
         # Replace last layer
         num_ftrs = self.network.fc.in_features
         self.network.fc = nn.Linear(num_ftrs, output_layer)
-        
+
     def forward(self, xb):
         x = torch.sigmoid(self.network(xb))
 
-        return x
+        return x