first commit

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+venv/

.gitignore

@@ -0,0 +1 @@
+venv/

README.md~

@@ -0,0 +1,17 @@
+# Pothole Detection App (Powered by YOLOv8 + Gradio)
+
+This is a lightweight web app that uses a custom-trained YOLOv8 model to detect potholes in uploaded images and videos. Built using Gradio, it provides a simple interface for end-users to test pothole detection instantly.
+
+## Features
+- 📷 Upload images for pothole detection.
+- 🎥 Upload short videos for pothole detection.
+- 🔊 Beep sound when a pothole is detected (in local mode).
+- 🌐 Can be hosted via platforms like Hugging Face Spaces or locally.
+
+## How to Run Locally
+
+```bash
+git clone https://github.com/Blazious/pothole_detection
+cd pothole_inference
+pip install -r requirements.txt
+python app.py

app.py

@@ -0,0 +1,47 @@
+# app.py
+import gradio as gr
+from detect import detect_image, detect_video
+
+def handle_image(img):
+    result_path, beep_path = detect_image(img)
+    return result_path, beep_path
+
+def handle_video(video):
+    result_path, beep_path = detect_video(video)
+    return result_path, beep_path
+
+with gr.Blocks() as demo:
+    gr.Markdown("## 🕳️ Pothole Detector")
+
+    with gr.Tabs():
+        with gr.TabItem("Image Detection"):
+            with gr.Row():
+                img_input = gr.Image(type="pil", label="Upload Image")
+                img_output = gr.Image(label="Detection Result")
+            audio_output_img = gr.Audio(label="Detection Beep", type="filepath", visible=True)
+            img_input.change(fn=handle_image, inputs=[img_input], outputs=[img_output, audio_output_img])
+
+        with gr.TabItem("Video Detection"):
+            with gr.Row():
+                video_input = gr.Video(label="Upload Video")
+                video_output = gr.Video(label="Detection Result")
+            audio_output_vid = gr.Audio(label="Detection Beep", type="filepath", visible=True)
+            video_input.change(fn=handle_video, inputs=[video_input], outputs=[video_output, audio_output_vid])
+
+        gr.Markdown("""
+                    ### 🔍 About This Tool
+
+                    - Designed for demonstration and educational purposes  
+                    - AI-generated pothole detection may not be 100% accurate  
+                    - Always verify road conditions with physical inspection  
+
+                    ---
+
+                    ⏱️ **Processing Note:**  
+                    Video analysis duration varies depending on the video's length and resolution.  
+                    Typical processing can take **5–6 minutes** as the system carefully analyzes individual frames.
+                    """)
+
+demo.launch()
+
+

detect.py

@@ -0,0 +1,75 @@
+# detect.py
+import os
+import cv2
+import torch
+from ultralytics import YOLO
+import tempfile
+from PIL import Image
+import numpy as np
+
+# Load model once
+model = YOLO('model/best.pt')
+
+# Load beep
+beep_path = os.path.join('media', 'beep.wav')
+
+def detect_image(img: Image.Image):
+    frame = np.array(img)
+
+    results = model.predict(source=frame, conf=0.4)
+    detections = results[0].boxes
+
+    detected = False
+
+    for box in detections:
+        cls_id = int(box.cls[0])
+        conf = float(box.conf[0])
+        x1, y1, x2, y2 = map(int, box.xyxy[0])
+        cv2.rectangle(frame, (x1, y1), (x2, y2), (0, 0, 255), 2)
+        label = f"Pothole: {conf:.2f}"
+        cv2.putText(frame, label, (x1, y1 - 10), cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 255, 0), 2)
+        detected = True
+
+    # Save to temp file
+    result_path = tempfile.mktemp(suffix=".jpg")
+    cv2.imwrite(result_path, frame)
+
+    return result_path, beep_path if detected else None
+
+# detect.py (for videos)
+def detect_video(video_path: str):
+    cap = cv2.VideoCapture(video_path)
+    fourcc = cv2.VideoWriter_fourcc(*'mp4v')
+    out_path = tempfile.mktemp(suffix=".mp4")
+    out = None
+    detected = False
+
+    while cap.isOpened():
+        ret, frame = cap.read()
+        if not ret:
+            break
+
+        results = model.predict(source=frame, conf=0.4)
+        boxes = results[0].boxes
+
+        for box in boxes:
+            cls_id = int(box.cls[0])
+            conf = float(box.conf[0])
+            x1, y1, x2, y2 = map(int, box.xyxy[0])
+            cv2.rectangle(frame, (x1, y1), (x2, y2), (0, 0, 255), 2)
+            label = f"Pothole: {conf:.2f}"
+            cv2.putText(frame, label, (x1, y1 - 10), cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 255, 0), 2)
+            detected = True
+
+        if out is None:
+            height, width, _ = frame.shape
+            out = cv2.VideoWriter(out_path, fourcc, 20.0, (width, height))
+
+        out.write(frame)
+
+    cap.release()
+    if out:
+        out.release()
+
+    return out_path, beep_path if detected else None
+

model/best.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9e9841997f0895771b6f2daa6f25560fc59b00905d7cb0ef7bf2622ed3c36f17
+size 6244387

requirements.txt

@@ -0,0 +1,6 @@
+ultralytics
+gradio
+opencv-python
+numpy
+Pillow
+moviepy