Scales

.gitignore

@@ -12,3 +12,8 @@ Menhaden Scales Aging/Inference Script/sam_vit_b_01ec64.pth
 Menhaden Scales Aging/Chapter 4 - Multimodal Train and Inference/best_model.pth
 Menhaden Scales Aging/Chapter 4 - Multimodal Train and Inference/final_model.pth
 Menhaden Scales Aging/Chapter 1 - Data Preprocessing/Raw Image/27061.tif
+
+.quarto/
+_site/
+
+Dockerfile

Menhaden Scales Aging/Inference Script/Scale_Aging_Inference_Script_Image_Only.py

@@ -1,12 +1,29 @@
+#!/usr/bin/env python
+
+# -----------------------------------------------------------------------------
+# Title: Scale_Aging_Inference_Script_Image_Only.py
+#
+# Description: This script predicts the age of fish scale images using a pre-
+#              trained ResNet18 model. Arguments, hyperparameters, and other
+#              settings are included in a configs.yml file. Predicted ages are
+#              written to a CSV file.
+#
+# Author: aotian.zheng@noaa.gov
+# Release Date: July 2025
+# Last Updated: August 2025
+#
+# Usage: python Scale_Aging_Inference_Script_Image_Only.py -c path/to/configs.yml
+# -----------------------------------------------------------------------------
+
 import argparse
+import yaml
 import cv2 as cv
-from matplotlib import pyplot as plt
 import numpy as np
 import os
 from os import listdir
 from os.path import isfile, join
-import numpy as np
 from PIL import Image
+import torch
 from torchvision import transforms
 from torchvision.io import read_image
 from torch.utils.data.dataset import Dataset  # For custom datasets
@@ -15,48 +32,68 @@
 from torch.utils.data import DataLoader
 from torchvision.models import resnet18, ResNet18_Weights
 from tqdm import tqdm
-import torch
-import yaml
 
 class FishTestDataset(Dataset):
+    """Custom Dataset for loading fish scale images for age inference.
+
+    Attributes
+    ----------
+    image_dir : str
+        Path to the directory containing images.
+    image_name : list
+        List of image filenames in the directory.
+    transforms : callable, optional
+        A function/transform that takes in a PIL image and returns a transformed version.
+
+    Methods
+    -------
+    __len__ : returns the number of images in the dataset.
+    __getitem__(index) : returns the image and its filename at the specified index.
+    """
     def __init__(self, image_dir, transform=None):
-
-        # Get the directory dataset images
+        """
+        Parameters
+        ----------
+        image_dir : str
+            Path to the directory containing images.
+        transform : callable, optional
+            A function/transform that takes in a PIL image and returns a transformed version.
+        """
+
+        # Get the directory of the images to age
         self.image_dir = image_dir
 
         # Get the transform methods
         self.transforms = transform
 
-
         # Image Name
         self.image_name = [f for f in listdir(image_dir) if isfile(join(image_dir, f))]
 
-
     def __len__(self):
+        """Returns the number of images in the dataset."""
         return len(self.image_name)
 
     def __getitem__(self, index):
+        """Returns the image and its filename at the specified index."""
+        # Open the specified image
         img_path = os.path.join(self.image_dir, str(self.image_name[index]))
         image = Image.open(img_path)
-
+        # Transform the image, if transforms are provided
         if self.transforms:
             image = self.transforms(image)
 
         return image, self.image_name[index]
 
-
+# Parse command line arguments. Currently only requires a path to a configuration yaml file.
 parser = argparse.ArgumentParser()
-parser.add_argument("--config_path", help="path to configuration yaml file")
-
+parser.add_argument("-c", "--config_path", help="path to configuration yaml file")
 args = parser.parse_args()
 
-        
+# Open the configuration file and read in the parameters
 with open(args.config_path, 'r') as file:
     config = yaml.safe_load(file)
 
-
-data_dir = config["img_path"]
-
+# Image transformations: resizing, cropping, normalization
 data_transforms = transforms.Compose(
         [
             transforms.Resize(224),
@@ -65,24 +102,22 @@ def __getitem__(self, index):
             transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225]),
         ]
     )
-test_dataset = FishTestDataset( data_dir, data_transforms)
+test_dataset = FishTestDataset(image_dir=config["image_path"], transform=data_transforms)
 test_loader = DataLoader(test_dataset, batch_size=24, shuffle=False, drop_last=False)
 
+# Load the model using GPU, if available, in evaluation mode.
+# Number of classes corresponds to the number of age classes.
 device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
-model = resnet18(num_classes = 5)
-
-# Load model - TODO
+model = resnet18(num_classes=5)
 model.load_state_dict(torch.load(config["model_path"]))
-
 model.eval()    
 model.to(device)
 
-import torch
-
-output_path = "inference_results.csv"
+# Create output file and write header
 file = open(config["out_path"], 'w')
 file.write("Image Name, Predicted Age\n")
 
+# Loop through the dataset and make predictions
 for images, img_path in tqdm(test_loader):
     images = images.to(device)
     outputs = model(images)
@@ -91,7 +126,8 @@ def __getitem__(self, index):
     preds = preds.cpu().detach().numpy()
     for i in range(preds.shape[0]):
         age = str(preds[i])
-        if(preds[i] ==4):
+        # Change the maximum age class to "4+"
+        if(preds[i] == 4):
             age = "4+"
-        file.write("%s,%s\n"%(img_path[i],age))
+        file.write("%s,%s\n" % (img_path[i], age))
 file.close()