Implement siamese network for one shot leraning

README.md

@@ -10,33 +10,45 @@ You can run one shot learning step by step. Also, I posted the details of the co
 
 ### 🚀How to run
 
-All executions begin at the location of `./siamese`. You can execute three action. *just run*, *download-data*, *train*, *test*.
+You can execute three action. *just run*, *download-data*, *train*, *test*.
 
-1. #### Run
+1. #### Clone
 
-   This commend automatically executes the entire process according to `config_maker`. If you just want to try this network, I recommend this.
+   Clone this repository and go into the directory.
+
+   ```bash
+   git clone https://github.com/Rhcsky/siamese-one-shot-pytorch.git
+
+   cd siamese-one-shot-pytorch
+   ```
+
+2. #### Run
+
+   This commend automatically executes the entire process according to `config_maker`(download data + train + test).
+
+   If you just want to try this network, I recommend this.
 
    ```bash
    python main.py run
    ```
 
-2. #### Download-data
+3. #### Download-data
 
    The Omniglot data is downloaded and divided into 30 types of train data, 10 types of validation data, and 10 types of test data. All data is contained in `./data/processed/`.
 
    ```bash
    python main.py download-data
    ```
 
-3. #### Train
+4. #### Train
 
    Only model learning is conducted. If you want to run 'train', you have to run 'download-data' first.
 
    ```bash
    python main.py train
    ```
 
-4. #### Test
+5. #### Test
 
    Only test the model. Stored models and datasets must exist.
 
@@ -48,7 +60,7 @@ All parameters are present in `config_maker`. If you want to adjust the paramete
 
 
 
-### Result
+### Check Result
 
 Train logs, saved model and configuration data were in `./result/[model_number]`.  Logs are made by `tensorboard`. So if you want to see more detail about train metrics, write commend on `./siamese_network/result/[model_number]`  like this.
 
@@ -60,8 +72,6 @@ tensorboard --logdir=logs
 
 ### 📌Reference
 
-siamese network
-
 * [Siamese Neural Networks for One-shot Image Recognition](https://www.cs.cmu.edu/~rsalakhu/papers/oneshot1.pdf)
 
 * [kevinzakka/one-shot-siamese](https://github.com/kevinzakka/one-shot-siamese)

siamese_network/data_prepare.py → data_prepare.py

@@ -6,7 +6,7 @@
 from tqdm import tqdm
 
 
-def copy_image_to_processed_dir(alpha_list, img_dir, desc):
+def move_image_to_processed_dir(alpha_list, img_dir, desc):
     for alpha in tqdm(alpha_list, desc=desc):
         write_dir1 = img_dir + '/' + os.path.basename(alpha) + '_'
         for char in (os.listdir(alpha)):
@@ -19,17 +19,21 @@ def copy_image_to_processed_dir(alpha_list, img_dir, desc):
 
 
 def prepare_data():
-    background_dir = "../data/unzip/background"
-    evaluation_dir = "../data/unzip/evaluation"
-    processed_dir = "../data/processed"
+    background_dir = "data/unzip/background"
+    evaluation_dir = "data/unzip/evaluation"
+    processed_dir = "data/processed"
     random.seed(5)
 
+    if not os.path.exists(processed_dir):
+        os.makedirs(processed_dir)
+        os.makedirs(processed_dir +'/train')
+        os.makedirs(processed_dir +'/val')
+        os.makedirs(processed_dir +'/test')
+
+
     if any([True for _ in os.scandir(processed_dir)]):
         return
 
-    if os.path.exists(processed_dir) is None:
-        os.makedirs(processed_dir)
-
     # Move 10 of evaluation image for getting more train set.
     if len(glob(evaluation_dir + '/*')) >= 20:
         for d in random.sample(glob(evaluation_dir + '/*'), 10):
@@ -48,6 +52,6 @@ def prepare_data():
     val_dir = os.path.join(processed_dir, 'val')
     test_dir = os.path.join(processed_dir, 'test')
 
-    copy_image_to_processed_dir(train_alpha, train_dir, 'train')
-    copy_image_to_processed_dir(val_alpha, val_dir, 'val')
-    copy_image_to_processed_dir(test_alpha, test_dir, 'test')
+    move_image_to_processed_dir(train_alpha, train_dir, 'train')
+    move_image_to_processed_dir(val_alpha, val_dir, 'val')
+    move_image_to_processed_dir(test_alpha, test_dir, 'test')

siamese_network/utils.py → utils.py

@@ -65,13 +65,15 @@ def load_config(config):
 
 # download omniglot dataset
 def download_omniglot_data():
-    BASEDIR = os.path.abspath(os.path.join(os.path.dirname(os.path.realpath(__file__)), os.pardir)) + '/data'
+    BASEDIR = os.path.dirname(os.path.realpath(__file__)) + "/data"
 
     # make directory
     if not os.path.exists(BASEDIR):
         os.mkdir(BASEDIR)
     if not os.path.exists(os.path.join(BASEDIR, 'unzip')):
         os.mkdir(os.path.join(BASEDIR, 'unzip'))
+    if not os.path.exists(os.path.join(BASEDIR, 'raw')):
+        os.mkdir(os.path.join(BASEDIR, 'raw'))
 
     # download zip file
     if not os.path.exists(BASEDIR + '/raw/images_background.zip'):