🐛 fix(app): add input degrees having a bug for cases without augmentation

- modify `predict_image_with_html` function to accept `augment_method` and `rotate_degrees`
- add filter to conditionally show rotate slider
- modify `augment_image` function to accept `methods` and `rotate_degrees` parameters and apply the chosen augmentation methods accordingly

✨ feat(app): expand augmentation methods

- add new checkbox group to let user select from multiple augmentation methods such as rotate

📝 docs: update imports comment and reorganize

- type hint comments for function return values
- add new filters to `PIL` Libraries for image manipulation

app.py

@@ -200,15 +200,15 @@ def generate_results_html(results):
     """
     return html_content
 
-def predict_image_with_html(img, confidence_threshold, augment):
-    if augment:
-        img_pil, _ = augment_image(img)
+def predict_image_with_html(img, confidence_threshold, augment_method, rotate_degrees):
+    if augment_method != "none":
+        img_pil, _ = augment_image(img, augment_method, rotate_degrees)
     else:
         img_pil = img
     img_pil, results = predict_image(img_pil, confidence_threshold)
     html_content = generate_results_html(results)
     return img_pil, html_content
- 
+
 with gr.Blocks() as iface:
     gr.Markdown("# AI Generated Image / Deepfake Detection Models Evaluation")
     
@@ -217,8 +217,9 @@ with gr.Blocks() as iface:
             image_input = gr.Image(label="Upload Image to Analyze", sources=['upload'], type='pil')
             with gr.Accordion("Settings", open=False, elem_id="settings_accordion"):
                 confidence_slider = gr.Slider(0.0, 1.0, value=0.5, step=0.01, label="Confidence Threshold")
-                augment_checkbox = gr.Checkbox(label="Augment Image", value=False)
-            inputs = [image_input, confidence_slider, augment_checkbox]
+                augment_checkboxgroup = gr.CheckboxGroup(["rotate", "add_noise", "sharpen"], label="Augmentation Methods")
+                rotate_slider = gr.Slider(0, 360, value=0, step=1, label="Rotate Degrees", visible=False)
+            inputs = [image_input, confidence_slider, augment_checkboxgroup, rotate_slider]
             predict_button = gr.Button("Predict")
         with gr.Column(scale=2):
             with gr.Accordion("Project OpenSight - Model Evaluations & Playground", open=False, elem_id="project_accordion"):
@@ -228,6 +229,8 @@ with gr.Blocks() as iface:
             results_html = gr.HTML(label="Model Predictions")
             outputs = [image_output, results_html]
     
+    augment_checkboxgroup.change(lambda methods: gr.Slider.update(visible="rotate" in methods), inputs=[augment_checkboxgroup], outputs=[rotate_slider])
+
     predict_button.click(
         fn=predict_image_with_html, 
         inputs=inputs, 

utils/utils.py

@@ -1,22 +1,22 @@
 import numpy as np
 import io
-from PIL import Image
+from PIL import Image, ImageFilter
 from torchvision import transforms
 
 def softmax(vector):
     e = np.exp(vector - np.max(vector))  # for numerical stability
     return e / e.sum()
 
-def augment_image(img_pil):
-    transform_flip = transforms.Compose([
-        transforms.RandomHorizontalFlip(p=1.0)
-    ])
-    transform_rotate = transforms.Compose([
-        transforms.RandomRotation(degrees=(90, 90))
-    ])
-    augmented_img_flip = transform_flip(img_pil)
-    augmented_img_rotate = transform_rotate(img_pil)
-    return augmented_img_flip, augmented_img_rotate
+def augment_image(img_pil, methods, rotate_degrees=0):
+    for method in methods:
+        if method == "rotate":
+            img_pil = img_pil.rotate(rotate_degrees)
+        elif method == "add_noise":
+            noise = np.random.normal(0, 25, img_pil.size[::-1] + (3,)).astype(np.uint8)
+            img_pil = Image.fromarray(np.clip(np.array(img_pil) + noise, 0, 255).astype(np.uint8))
+        elif method == "sharpen":
+            img_pil = img_pil.filter(ImageFilter.SHARPEN)
+    return img_pil, img_pil
 
 def convert_pil_to_bytes(image, format='JPEG'):
     img_byte_arr = io.BytesIO()