minor fix for v1.2

Dockerfile

@@ -23,7 +23,7 @@ RUN git clone https://github.com/Kawaeee/butt_or_bread.git
 
 # Download model
 RUN cd /opt/butt_or_bread/
-RUN wget https://github.com/Kawaeee/butt_or_bread/releases/download/v1.1/buttbread_resnet152_3.h5
+RUN wget https://github.com/Kawaeee/butt_or_bread/releases/download/v1.2/buttbread_resnet152_3.h5
 
 # Install python packages
 RUN pip install --upgrade pip

README.md

@@ -18,7 +18,7 @@
 
   **After that, we manually remove incorrect images and apply [phash](https://github.com/Kawaeee/phash) (Perceptual Hashing) to get rid of duplication images.**
 
-  * Total images: 6385 images (randomly split using 80:10:10 ratios) with [dataset-split](https://github.com/muriloxyz/dataset-split)
+  * Total images: 6385 images (randomly split using 80:10:10 ratios)
 
     * **Bread**: 3710 images
       * Train: 2968 images
@@ -39,9 +39,9 @@
 |**Valid**|0.0132|0.9969|
 |**Test**|-|0.9968|
 
-* We already know that to benchmark our model performance, we can't just use `accuracy` and `validation_loss` value as the only acceptable metrics.
+* We already know that in order to benchmark our model performance, we can't just use `accuracy` and `validation_loss` value as the only acceptable metrics.
 
-#### You can download our model weight here: [v1.2](https://github.com/Kawaeee/butt_or_bread/releases/download/v1.3/buttbread_resnet152_3.h5)
+#### You can download our model weight here: [v1.2](https://github.com/Kawaeee/butt_or_bread/releases/download/v1.2/buttbread_resnet152_3.h5)
 
 ## Hyperparameters and configurations
 
@@ -54,7 +54,7 @@
 ## Dataset Preparation
  * To reproduce the model, requires our datasets. You can send me an e-mail at `[email protected]` if you are interested.
 
- - Prepare dataset in these following directory structure
+ - Initial datasets/ directory structure
    ```Bash
       └───datasets/
       │     butt/
@@ -66,12 +66,12 @@
    pip install dataset-split
    ```
 
- - Execute `dataset-split` command with following arguments on both category
+ - Execute `dataset-split` command with following arguments
    ```bash
    dataset-split dataset/ -r 0.8 0.1 0.1
    ```
 
--  The result will be in this format, and we are ready to proceed to model training
+-  Ready-to-go datasets/ directory structure
    ```Bash
       └───datasets/
       │   │
@@ -136,5 +136,3 @@
    ```Bash
    docker run -p 8501:8501 butt_or_bread
    ```
-
- - Streamlit web application will be hosted on http://localhost:8501

streamlit_app.json

@@ -4,7 +4,7 @@
       "icon":"assets/icon/corgi-icon.png",
       "markdown":{
          "star":"[![GitHub Star](https://img.shields.io/github/stars/Kawaeee/butt_or_bread)](https://github.com/Kawaeee/butt_or_bread)",
-         "release":"[![GitHub Release](https://img.shields.io/github/v/release/Kawaeee/butt_or_bread)](https://github.com/Kawaeee/butt_or_bread/releases/tag/v1.1)",
+         "release":"[![GitHub Release](https://img.shields.io/github/v/release/Kawaeee/butt_or_bread)](https://github.com/Kawaeee/butt_or_bread/releases/tag/v1.2)",
          "visitor":"![Visitor Badge](https://visitor-badge.glitch.me/badge?page_id=Kawaeee.butt_or_bread.visitor-badge)"
       },
       "mode":{
@@ -20,7 +20,7 @@
       }
    },
    "model":{
-      "url":"https://github.com/Kawaeee/butt_or_bread/releases/download/v1.1/buttbread_resnet152_3.h5",
+      "url":"https://github.com/Kawaeee/butt_or_bread/releases/download/v1.2/buttbread_resnet152_3.h5",
       "label":{
          "corgi":"Corgi butt 🐕",
          "bread":"Loaf of bread 🍞"

train.py

@@ -10,17 +10,20 @@
 
 
 class ButtBreadModel:
+    """Corgi butt or loaf of bread? model"""
+
     def __init__(self, device):
         self.model = None
         self.device = device
         self.criterion = None
         self.optimizer = None
 
     def initialize(self):
+        """Transfer Learning by using ResNet-152 as pre-trained weight"""
         self.model = models.resnet152(pretrained=True).to(self.device)
 
-        for param in self.model.parameters():
-            param.requires_grad = False
+        for parameter in self.model.parameters():
+            parameter.requires_grad = False
 
         self.model.fc = torch.nn.Sequential(
             torch.nn.Linear(2048, 128),
@@ -47,7 +50,7 @@ def train(self, image_dataloaders, image_datasets, epochs=1):
                 running_loss = 0.0
                 running_corrects = 0
 
-                # Iterate and try to predict input and check with output  generate loss and correct label
+                # Iterate and try to predict input and check with output -> generate loss and correct label
                 for inputs, labels in tqdm(image_dataloaders[phase]):
                     inputs = inputs.to(self.device)
                     labels = labels.to(self.device)
@@ -65,25 +68,24 @@ def train(self, image_dataloaders, image_datasets, epochs=1):
                     running_corrects += torch.sum(preds == labels.data)
 
                 epoch_loss = running_loss / len(image_datasets[phase])
-                epoch_acc = running_corrects.float() / len(image_datasets[phase])
+                epoch_accuracy = running_corrects.float() / len(image_datasets[phase])
 
-                print(f"{phase} loss: {epoch_loss.item():.4f}, acc: {epoch_acc.item():.4f}")
+                print(f"{phase} loss: {epoch_loss.item():.4f}, acc: {epoch_accuracy.item():.4f}")
 
             print("Runtime: (", "{0:.2f}".format(time.monotonic() - time_start), " seconds)", sep="")
 
         return self.model
 
     def test(self, image_dataloaders):
         """Test with test set"""
-        test_acc_count = 0
+        test_accuracy_count = 0
 
         for k, (test_images, test_labels) in tqdm(enumerate(image_dataloaders["test"])):
             test_outputs = self.model(test_images.to(self.device))
             _, prediction = torch.max(test_outputs.data, 1)
-            test_acc_count += torch.sum(prediction == test_labels.to(self.device).data).item()
+            test_accuracy_count += torch.sum(prediction == test_labels.to(self.device).data).item()
 
-        test_accuracy = test_acc_count / len(image_dataloaders["test"])
-        print(f"Test acc: {test_accuracy}")
+        test_accuracy = test_accuracy_count / len(image_dataloaders["test"])
 
         return test_accuracy
 
@@ -168,8 +170,11 @@ def main(opt):
         epochs=epochs,
     )
 
-    butt_bread_obj.test(image_dataloaders=image_dataloaders)
+    test_accuracy = butt_bread_obj.test(image_dataloaders=image_dataloaders)
+    print(f"Test accuracy: {test_accuracy}")
+
     butt_bread_obj.save(model_path=model_path)
+    print(f"Saved model at {model_path}")
 
 
 if __name__ == "__main__":