code cleaned

.gitignore

@@ -1 +1,2 @@
 venv
+.DS_Store

README.md

@@ -10,3 +10,31 @@ pinned: false
 ---
 
 Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+
+
+The dataset used during the training is also uploaded to HUgging Face here: https://huggingface.co/datasets/trifork/plastic-pellets
+
+
+
+
+To successfully push binary image files or models to your Hugging Face repository, you'll need to use Git Large File Storage (Git LFS). Git LFS is designed to handle large files by storing them separately from your main Git repository, which prevents issues like the one you're encountering. To do that you have to:
+
+1. Install Git LFS
+'''
+brew install git-lfs
+'''
+
+2. Initialize Git LFS in Your Repository
+'''
+git lfs install
+'''
+
+3. Track the Binary File Types, for example:
+'''
+git lfs track "*.bmp"
+'''
+4. Add .gitattributes to Git
+'''
+git add .gitattributes
+git commit -m "Configure Git LFS to track .bmp files"
+'''

app.py

@@ -1,163 +1,89 @@
 # app.py
+# app.py
 import gradio as gr
 from PIL import Image, ImageDraw, ImageFont
 import torch
 from ultralytics import YOLO
 import numpy as np
 import os
-
-# Print Pillow version for debugging
 from PIL import __version__ as PIL_VERSION
 print(f"Pillow version: {PIL_VERSION}")
 
-# Define the path to your custom YOLO model
-MODEL_PATH = "model/231220_detect_lr_0001_640_brightness.pt"  # Ensure this path is correct
+MODEL_PATH = "model/231220_detect_lr_0001_640_brightness.pt"
 
-# Verify that the model file exists
+# Verify the model path
 if not os.path.exists(MODEL_PATH):
-    raise FileNotFoundError(
-        f"YOLO model not found at '{MODEL_PATH}'. Please ensure the model file is in the 'model/' directory."
-    )
+    raise FileNotFoundError(f"YOLO model not found at '{MODEL_PATH}'.")
 
-# Initialize the YOLO model
-print("Loading YOLO model...")
+# Load the YOLO model
 model = YOLO(MODEL_PATH)
-print("YOLO model loaded successfully.")
+print("YOLO model loaded.")
 
 def detect_plastic_pellets(input_image):
     """
-    Perform plastic pellet detection using the YOLO model.
-
-    Args:
-        input_image (PIL.Image): The input beach image uploaded by the user.
-
-    Returns:
-        PIL.Image: The image with detected plastic pellets highlighted or an image with an error message.
+    Perform plastic pellet detection using our customized model.
     """
     if input_image is None:
-        print("No image uploaded.")
-        # Create an error image with a message
         error_image = Image.new('RGB', (500, 100), color=(255, 0, 0))
         draw = ImageDraw.Draw(error_image)
         try:
             font = ImageFont.truetype("arial.ttf", size=15)
         except IOError:
             font = ImageFont.load_default()
-        message = "Please upload a valid image."
-        bbox = font.getbbox(message)
-        draw.text((10, 40), message, fill=(255, 255, 255), font=font)
+        draw.text((10, 40), "Please upload a valid image.", fill=(255, 255, 255), font=font)
         return error_image
 
     try:
         print("Starting detection...")
-        # Optionally, resize the image to a maximum size to optimize performance
-        max_size = (1024, 1024)
-
-        # Attempt to use Resampling if available, else use LANCZOS
-        try:
-            from PIL import Resampling
-            resample_filter = Resampling.LANCZOS
-        except ImportError:
-            resample_filter = Image.LANCZOS  # For Pillow versions <10.0.0
-
-        input_image.thumbnail(max_size, resample_filter)
-        print(f"Image resized to: {input_image.size}")
-
-        # Convert PIL Image to numpy array
+        input_image.thumbnail((1024, 1024), Image.LANCZOS)
         img = np.array(input_image.convert("RGB"))
 
-        # Perform inference
         results = model(img)
-        print(f"Number of results: {len(results)}")
-
-        # Initialize drawing context
         draw = ImageDraw.Draw(input_image)
         try:
-            # Attempt to load a TrueType font
             font = ImageFont.truetype("arial.ttf", size=15)
         except IOError:
-            # Fall back to the default font if arial.ttf is not found
             font = ImageFont.load_default()
 
-        # Initialize a flag to check if any detection was made
         detection_made = False
 
-        # Iterate over detections and draw bounding boxes
         for result in results:
-            boxes = result.boxes
-            print(f"Number of boxes in result: {len(boxes)}")
-            for box in boxes:
-                # Extract bounding box coordinates and confidence
-                x1, y1, x2, y2 = box.xyxy[0].tolist()
+            for box in result.boxes:
+                x1, y1, x2, y2 = map(int, box.xyxy[0].tolist())
                 confidence = box.conf[0].item()
                 cls = int(box.cls[0].item())
                 name = model.names[cls] if model.names else "Object"
 
-                print(f"Detected object: {name}, Confidence: {confidence:.2f}")
-
-                # Define bounding box color and thickness
-                color = (255, 0, 0)  # Red color
-                thickness = 2
+                color = (255, 0, 0)
+                draw.rectangle(((x1, y1), (x2, y2)), outline=color, width=2)
 
-                # Convert coordinates to integers
-                x1, y1, x2, y2 = map(int, [x1, y1, x2, y2])
-
-                # Draw rectangle
-                draw.rectangle(((x1, y1), (x2, y2)), outline=color, width=thickness)
-
-                # Prepare label
                 label = f"{name} {confidence:.2f}"
+                text_width, text_height = font.getbbox(label)[2:]
+                draw.rectangle(((x1, y1 - text_height), (x1 + text_width, y1)), fill=color)
+                draw.text((x1, y1 - text_height), label, fill=(255, 255, 255), font=font)
 
-                # Calculate text size using font.getbbox
-                bbox = font.getbbox(label)
-                text_width = bbox[2] - bbox[0]
-                text_height = bbox[3] - bbox[1]
-
-                # Draw label background
-                draw.rectangle(
-                    ((x1, y1 - text_height), (x1 + text_width, y1)),
-                    fill=color
-                )
-
-                # Draw label text
-                draw.text(
-                    (x1, y1 - text_height),
-                    label,
-                    fill=(255, 255, 255),
-                    font=font
-                )
-
-                # Set the flag to True as a detection was made
                 detection_made = True
 
         if not detection_made:
-            print("No detections made.")
-            # Optionally, return the original image with a message drawn on it
-            message = "No plastic pellets detected."
-            bbox = font.getbbox(message)
-            draw.text((10, 10), message, fill=(255, 0, 0), font=font)
-            return input_image  # Still return an image, but with a message
+            draw.text((10, 10), "No plastic pellets detected.", fill=(255, 0, 0), font=font)
+            return input_image  
 
         print("Detection completed.")
         return input_image
 
     except Exception as e:
         print(f"Detection error: {str(e)}")
-        # Instead of returning a string, create an image with the error message
         error_image = Image.new('RGB', (500, 100), color=(255, 0, 0))
         draw = ImageDraw.Draw(error_image)
         try:
             font = ImageFont.truetype("arial.ttf", size=15)
         except IOError:
             font = ImageFont.load_default()
-        message = f"Error: {str(e)}"
-        bbox = font.getbbox(message)
-        draw.text((10, 40), message, fill=(255, 255, 255), font=font)
+        draw.text((10, 40), f"Error: {str(e)}", fill=(255, 255, 255), font=font)
         return error_image
 
 def main():
     with gr.Blocks(css=".gradio-container {max-width: 800px}") as demo:
-        # Header Section
         gr.Markdown(
             """
             <h1 align="center">🌊 Beach Plastic Pellet Detection Challenge</h1>
@@ -165,63 +91,23 @@ def main():
             """
         )
 
-        # Instructions Section
-        gr.Markdown(
-            """
-            ### 📸 How to Participate
-            1. **Upload or Select Image**: Click on the upload area to select a beach photo from your device or choose one of our sample images below.
-            2. **Detect**: Click the "Detect Plastic Pellets" button to run the model.
-            3. **Contribute**: Download the result and share it to help us collect data on plastic pellet pollution.
-            
-            ### 🌍 Why Participate?
-            By participating, you're helping us collect valuable data on plastic pellet pollution on beaches worldwide. Your contributions will aid in environmental research and clean-up initiatives, making a tangible impact on our planet's health.
-            """
-        )
-
-        # Upload and Output Section
         with gr.Row():
             with gr.Column():
-                input_image = gr.Image(
-                    type="pil",
-                    label="🌊 Upload or Select Beach Image",
-                    interactive=True,
-                )
-
-                # Examples Section
-                examples = [
-                    'images/image1.bmp',
-                    'images/image2.bmp',
-                    'images/image3.bmp'
-                ]
-
-                gr.Examples(
-                    examples=examples,
-                    inputs=input_image,
-                    label="Or choose one of these images",
-                    examples_per_page=3,
-                )
-
+                input_image = gr.Image(type="pil", label="🌊 Upload or Select Beach Image", interactive=True)
+                examples = ['images/image1.bmp', 'images/image2.bmp', 'images/image3.bmp']
+                gr.Examples(examples=examples, inputs=input_image, label="Or choose one of these images")
                 submit_button = gr.Button("🔍 Detect Plastic Pellets")
 
             with gr.Column():
-                output_image = gr.Image(
-                    type="pil",
-                    label="✅ Detection Result",
-                    interactive=False,
-                    show_download_button=True  # Add download button directly to the output image
-                )
-
-        # Footer Section
+                output_image = gr.Image(type="pil", label="✅ Detection Result", interactive=False, show_download_button=True)
+
         gr.Markdown(
             """
             ---
-            <p align="center">
-                © 2024 Beach Clean-Up Initiative.
-            </p>
+            <p align="center">© 2024 Beach Clean-Up Initiative.</p>
             """
         )
 
-        # Connect the button to the detection function
         submit_button.click(
             fn=detect_plastic_pellets,
             inputs=input_image,