Refactor app.py to remove model caching and initialize model per request. Update root endpoint and modify predict function to handle uploaded images. Change test.py to use variable for local app URL.

app.py

@@ -1,22 +1,10 @@
 from fastapi import FastAPI, UploadFile
 from ultralytics import YOLO
-import io
 from PIL import Image
-import numpy as np
 import os
 from huggingface_hub import hf_hub_download
-from ultralytics import YOLO
-import requests
-import supervision as sv
-
-# Global variable to store model instances
-MODEL_CACHE = {}
 
 def init_model(model_id: str):
-    # Return cached model if it exists
-    if model_id in MODEL_CACHE:
-        return MODEL_CACHE[model_id]
-    
     # Define models
     MODEL_OPTIONS = {
         "YOLOv11-Nano": "medieval-yolov11n.pt",
@@ -28,34 +16,33 @@ def init_model(model_id: str):
     
     if model_id in MODEL_OPTIONS:
         os.makedirs("models", exist_ok=True)
-        model_path = hf_hub_download(
+        path = hf_hub_download(
             repo_id="biglam/medieval-manuscript-yolov11",
-            filename=MODEL_OPTIONS[model_id],
-            cache_dir="models"  # Specify cache directory
+            filename=MODEL_OPTIONS[model_id]
         )
-        model = YOLO(model_path)
-        MODEL_CACHE[model_id] = model
+        local_path = os.path.join("models", path)
+        # Initialize and return model
+        model = YOLO(path)
         return model
     else:
         raise ValueError(f"Model {model_id} not found")
 
 app = FastAPI()
 
-# Initialize default model at startup
-@app.on_event("startup")
-async def startup_event():
-    init_model("YOLOv11-XLarge")  # Initialize default model
+@app.get("/")
+async def root():
+    return {"status": "ok"}
 
 @app.post("/predict")
 async def predict(image: UploadFile,
-                  model_id: str = "YOLOv11-XLarge",
-                  conf: float = 0.25,
-                  iou: float = 0.7
-                  ):
-    # Get model from cache or initialize it
+                 model_id: str = "YOLOv11-XLarge",
+                 conf: float = 0.25,
+                 iou: float = 0.7
+                 ):
+    # Initialize model for each request
     model = init_model(model_id)
     
-    # Download and open image from URL
+    # Open image from uploaded file
     image = Image.open(image.file)
     
     # Run inference with the PIL Image
@@ -63,18 +50,16 @@ async def predict(image: UploadFile,
     
     # Extract detection results
     result = results[0]
-    # print(result)
     detections = []
     
     for box in result.boxes:
         detection = {
             "class": result.names[int(box.cls[0])],
             "confidence": float(box.conf[0]),
-            "bbox": box.xyxy[0].tolist()  # Convert bbox tensor to list
+            "bbox": box.xyxy[0].tolist()
         }
         detections.append(detection)
-    print(detections)
-
+    
     return {"detections": detections}
 
 if __name__ == "__main__":

test.py

@@ -18,7 +18,7 @@ with open(image_path, 'rb') as f:
     }
 
     # Send POST request to the endpoint
-    response = requests.post(hf_app + ':7860/predict', 
+    response = requests.post(local_app + ':7860/predict', 
                            files=files,
                            params=params)
 

test2.py

@@ -0,0 +1,21 @@
+from concurrent.futures import ThreadPoolExecutor
+
+def process_item(item):
+    # Simulate a time-consuming task
+    import time
+    time.sleep(0.1)
+    return item * 2
+
+def generate_data():
+    for i in range(10):
+        yield i
+
+def process_data_in_parallel(data_generator):
+    with ThreadPoolExecutor(max_workers=4) as executor:
+        results = executor.map(process_item, data_generator)
+        return list(results)
+
+# Usage example
+data_generator = generate_data()
+processed_data = process_data_in_parallel(data_generator)
+print(processed_data)