Upload 3 files

building YOLO12s space

best.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8f29aea1f80a5c713ca2d8094d751df7097620d0a29835b3729eddc58ed297c1
+size 18914643

main.py

@@ -0,0 +1,168 @@
+import gradio as gr
+import mlflow
+import dagshub
+from ultralytics import YOLO
+from PIL import Image
+import cv2
+import numpy as np
+import os
+import time
+
+# Initialize DagsHub/MLflow
+dagshub.init(repo_owner='Mosensei7', repo_name='AutonomousVehiclesDetectionDEPI', mlflow=True)
+mlflow.set_experiment("Inference_Experiments")
+
+# Load your trained model
+model = YOLO('best.pt')  # Replace with your best.pt path
+
+def inference(media, media_type):
+    with mlflow.start_run(run_name=f"Inference_{int(time.time())}") as run:
+        # Log parameters
+        mlflow.log_param("media_type", media_type)
+        mlflow.log_param("timestamp", time.time())
+        
+        if media_type == "Image":
+            # Process image
+            results = model(media)[0]
+            output = results.plot()
+            output_pil = Image.fromarray(output)
+            
+            # Log input/output as artifacts
+            input_path = "input_image.jpg"
+            output_path = "output_image.jpg"
+            media.save(input_path)
+            output_pil.save(output_path)
+            mlflow.log_artifact(input_path)
+            mlflow.log_artifact(output_path)
+            
+            # Log metrics if available
+            if hasattr(results, 'boxes'):
+                num_detections = len(results.boxes)
+                mlflow.log_metric("num_detections", num_detections)
+            
+            os.remove(input_path)
+            os.remove(output_path)
+            
+            return output_pil, run.info.run_id
+        
+        elif media_type == "Video":
+            # Process video
+            cap = cv2.VideoCapture(media)
+            fourcc = cv2.VideoWriter_fourcc(*'mp4v')
+            out = cv2.VideoWriter('output.mp4', fourcc, cap.get(cv2.CAP_PROP_FPS), 
+                                  (int(cap.get(3)), int(cap.get(4))))
+            
+            frame_count = 0
+            detections_count = 0
+            
+            while cap.isOpened():
+                ret, frame = cap.read()
+                if not ret:
+                    break
+                    
+                results = model(frame)[0]
+                annotated = results.plot()
+                out.write(annotated)
+                
+                frame_count += 1
+                detections_count += len(results.boxes) if hasattr(results, 'boxes') else 0
+            
+            cap.release()
+            out.release()
+            
+            # Log video artifacts
+            mlflow.log_artifact(media, "input_video.mp4")
+            mlflow.log_artifact("output.mp4", "output_video.mp4")
+            
+            # Log metrics
+            mlflow.log_metric("frame_count", frame_count)
+            mlflow.log_metric("total_detections", detections_count)
+            
+            os.remove("output.mp4")
+            
+            return "output.mp4", run.info.run_id
+
+    return None, None
+
+# Futuristic CSS
+css = """
+body {
+    background: linear-gradient(135deg, #0f0c29, #302b63, #24243e);
+    color: #ffffff;
+    font-family: 'Orbitron', sans-serif;
+}
+
+.gradio-container {
+    background: rgba(0,0,0,0.3);
+    border: 2px solid #00ffff;
+    border-radius: 20px;
+    padding: 20px;
+    box-shadow: 0 0 20px #00ffff;
+}
+
+h1 {
+    text-align: center;
+    color: #00ffff;
+    text-shadow: 0 0 10px #00ffff;
+}
+
+button {
+    background: linear-gradient(45deg, #ff00ff, #00ffff);
+    border: none;
+    color: black;
+    font-weight: bold;
+    border-radius: 10px;
+    box-shadow: 0 0 10px #00ffff;
+}
+
+.input-area, .output-area {
+    border: 1px solid #00ffff;
+    border-radius: 10px;
+    padding: 10px;
+    background: rgba(255,255,255,0.05);
+}
+
+#mlflow-link {
+    color: #00ffff;
+    text-decoration: none;
+}
+"""
+
+with gr.Blocks(css=css) as demo:
+    gr.Markdown("""
+    <h1>🚀 Futuristic Vehicle Detection Interface</h1>
+    <p style="text-align: center; color: #00ffff; text-shadow: 0 0 5px #00ffff;">
+        Upload image or video for detection. Every inference logged to MLflow!
+    </p>
+    """)
+    
+    with gr.Row():
+        with gr.Column(scale=1, variant="panel", elem_classes="input-area"):
+            media = gr.File(label="Upload Image/Video", file_types=[".jpg", ".png", ".mp4"])
+            media_type = gr.Radio(["Image", "Video"], label="Media Type", value="Image")
+            submit = gr.Button("Detect Objects")
+        
+        with gr.Column(scale=2, variant="panel", elem_classes="output-area"):
+            output = gr.Image(label="Detection Result", type="pil")  # For image
+            video_output = gr.Video(label="Detection Result")  # For video
+            mlflow_id = gr.Textbox(label="MLflow Run ID (click to view log)")
+    
+    def process(media_path, m_type):
+        if m_type == "Image":
+            img = Image.open(media_path)
+            out_img, run_id = inference(img, m_type)
+            return out_img, None, run_id
+        else:
+            out_video, run_id = inference(media_path, m_type)
+            return None, out_video, run_id
+    
+    def show_mlflow_link(run_id):
+        if run_id:
+            url = f"https://dagshub.com/Mosensei7/AutonomousVehiclesDetectionDEPI/mlflow/runs/{run_id}"
+            return f"<a id='mlflow-link' href='{url}' target='_blank'>View Inference Log on DagsHub</a>"
+        return ""
+    
+    submit.click(process, [media, media_type], [output, video_output, mlflow_id])
+    mlflow_id.change(show_mlflow_link, mlflow_id, mlflow_id, show_label=False)
+
+demo.launch(share=True)

requirments.txt

@@ -0,0 +1,9 @@
+ultralytics>=8.3.0
+gradio>=4.0
+mlflow>=2.0
+dagshub>=0.3.0
+opencv-python>=4.8.0
+torch>=2.0.0
+torchvision>=0.15.0
+Pillow>=9.0.0
+numpy>=1.21.0