Upload 4 files

best.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a3899264fa4f4f1dd5b79b2bd748597aa7bb400e5ec57244b5dace63bb451c45
+size 52028609

packages.txt

@@ -0,0 +1 @@
+libgl1

requirements.txt

@@ -0,0 +1,11 @@
+Flask==3.0.3
+ipython==8.12.3
+numpy==2.0.0
+opencv_python==4.9.0.80
+opencv_python_headless==4.10.0.84
+pandas==2.2.2
+Pillow==10.4.0
+Requests==2.32.3
+roboflow==1.1.34
+streamlit==1.36.0
+ultralytics==8.0.196

streamlitTEST.py

@@ -0,0 +1,84 @@
+import streamlit as st
+from ultralytics import YOLO
+from PIL import Image, ImageDraw
+import tempfile
+
+def greeting(name):
+    return f"Hello from module2, {name}!"
+
+# Function to process the image, classify it, and crop if clear
+def process_image(file_path):
+    model = YOLO("best.pt", "v8")
+
+    # Predict with the model
+    with tempfile.NamedTemporaryFile(delete=False, suffix=".png") as temp_image:
+        temp_image.close()
+        Image.open(file_path).save(temp_image.name)
+        results = model.predict(source=temp_image.name, conf=0.4, save=True)
+
+    blur_conf_threshold = 0.5
+    clear_conf_threshold = 0.9
+
+    # Initialize flags
+    is_blur = False
+    is_clear = False
+    cropped_image = None
+
+    # Process results
+    for result in results[0].boxes:
+        confidence = result.conf[0].item()  # Extract the confidence score
+        if blur_conf_threshold <= confidence <= clear_conf_threshold:
+            is_blur = True
+        elif confidence > clear_conf_threshold:
+            is_clear = True
+            box = result.xyxy[0].tolist()  # Extract bounding box coordinates
+            cropped_image = crop_image(file_path, box)
+
+    # Return classification and cropped image
+    if is_blur:
+        return 'The image is blurry. Please reupload the image again!', None
+    elif is_clear:
+        return 'The image is clear', cropped_image
+    else:
+        s=("Not Detected! The image is uncertain. Please reupload the image again!")
+        return s, None
+
+# Function to crop the image based on bounding box
+def crop_image(file_path, box):
+    image = Image.open(file_path)
+    cropped_image = image.crop(box)
+    return cropped_image
+
+def greeting(name):
+    return f"Hello module2, {name}!"
+
+# Streamlit app
+def main():
+    st.title('Welcome to my AI project')
+    st.title('Document Detection')
+    st.text('This is a web app to:\n1- Detect documents\n2- Classify if document is clear or blurry\n3- Crop the document image!')
+
+    # File uploader
+    uploaded_file = st.file_uploader('Upload your image here:', type=['png', 'jpg', 'jpeg'])
+   
+    
+
+    if uploaded_file is not None:
+        # Display the uploaded image
+        image = Image.open(uploaded_file)
+        st.image(image, caption='Uploaded Image.', use_column_width=True)
+        st.success("Photo uploaded successfully!")
+
+        # Process the image and get classification and cropped image
+        classification, cropped_image = process_image(uploaded_file)
+
+        # Display classification result
+        st.write('Classification result:', classification)
+
+        # Display cropped image if classification is clear
+        if cropped_image is not None:
+            st.image(cropped_image, caption='Cropped Document Image.', use_column_width=True)
+
+if __name__ == "__main__":
+    main()
+