Update main.py

main.py

@@ -1,76 +1,71 @@
 from fastapi import FastAPI, File, UploadFile, HTTPException
-from fastapi.responses import JSONResponse
 from pydantic import BaseModel
-from typing import Optional
 import base64
 import io
+import os
+import logging
 from PIL import Image
 import torch
-import numpy as np
-import os
 
 # Existing imports
-import numpy as np
-import torch
-from PIL import Image
-import io
-
 from utils import (
     check_ocr_box,
     get_yolo_model,
     get_caption_model_processor,
     get_som_labeled_img,
 )
-import torch
-
-# yolo_model = get_yolo_model(model_path='/data/icon_detect/best.pt')
-# caption_model_processor = get_caption_model_processor(model_name="florence2", model_name_or_path="/data/icon_caption_florence")
-
-from ultralytics import YOLO
+from transformers import AutoProcessor, AutoModelForCausalLM
 
-# if not os.path.exists("/data/icon_detect"):
-#     os.makedirs("/data/icon_detect")
+# Configure logging
+logging.basicConfig(level=logging.DEBUG)  # Changed to DEBUG for more verbosity
+logger = logging.getLogger(__name__)
 
-try:
-    yolo_model = YOLO("weights/best.pt").to("cuda")
-except:
-    yolo_model = YOLO("weights/best.pt")
+# Load YOLO model
+yolo_model = get_yolo_model(model_path="weights/best.pt")
 
-from transformers import AutoProcessor, AutoModelForCausalLM
-
-processor = AutoProcessor.from_pretrained(
-    "microsoft/Florence-2-base", trust_remote_code=True
-)
+# Handle device placement
+device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+if str(device) == "cuda":
+    yolo_model = yolo_model.cuda()
+else:
+    yolo_model = yolo_model.cpu()
 
+# Load caption model and processor
 try:
+    processor = AutoProcessor.from_pretrained(
+        "microsoft/Florence-2-base", trust_remote_code=True
+    )
     model = AutoModelForCausalLM.from_pretrained(
         "weights/icon_caption_florence",
         torch_dtype=torch.float16,
         trust_remote_code=True,
     ).to("cuda")
-except:
+except Exception as e:
+    logger.warning(f"Failed to load caption model on GPU: {e}. Falling back to CPU.")
     model = AutoModelForCausalLM.from_pretrained(
         "weights/icon_caption_florence",
         torch_dtype=torch.float16,
         trust_remote_code=True,
     )
+
 caption_model_processor = {"processor": processor, "model": model}
-print("finish loading model!!!")
+logger.info("Finished loading models!!!")
 
 app = FastAPI()
 
-
 class ProcessResponse(BaseModel):
     image: str  # Base64 encoded image
     parsed_content_list: str
     label_coordinates: str
 
-
-def process(
-    image_input: Image.Image, box_threshold: float, iou_threshold: float
-) -> ProcessResponse:
+def process(image_input: Image.Image, box_threshold: float, iou_threshold: float) -> ProcessResponse:
     image_save_path = "imgs/saved_image_demo.png"
+    os.makedirs(os.path.dirname(image_save_path), exist_ok=True)
     image_input.save(image_save_path)
+
+    logger.info(f"Saved image for processing: {image_save_path}")
+
+    # Open image and prepare it for further processing
     image = Image.open(image_save_path)
     box_overlay_ratio = image.size[0] / 3200
     draw_bbox_config = {
@@ -80,6 +75,7 @@ def process(
         "thickness": max(int(3 * box_overlay_ratio), 1),
     }
 
+    # OCR and YOLO box processing
     ocr_bbox_rslt, is_goal_filtered = check_ocr_box(
         image_save_path,
         display_img=False,
@@ -89,33 +85,40 @@ def process(
         use_paddleocr=True,
     )
     text, ocr_bbox = ocr_bbox_rslt
-    dino_labled_img, label_coordinates, parsed_content_list = get_som_labeled_img(
-        image_save_path,
-        yolo_model,
-        BOX_TRESHOLD=box_threshold,
-        output_coord_in_ratio=True,
-        ocr_bbox=ocr_bbox,
-        draw_bbox_config=draw_bbox_config,
-        caption_model_processor=caption_model_processor,
-        ocr_text=text,
-        iou_threshold=iou_threshold,
-    )
+
+    # Process image and get result
+    try:
+        dino_labled_img, label_coordinates, parsed_content_list = get_som_labeled_img(
+            image_save_path,
+            yolo_model,
+            BOX_TRESHOLD=box_threshold,
+            output_coord_in_ratio=True,
+            ocr_bbox=ocr_bbox,
+            draw_bbox_config=draw_bbox_config,
+            caption_model_processor=caption_model_processor,
+            ocr_text=text,
+            iou_threshold=iou_threshold,
+        )
+    except Exception as e:
+        logger.error(f"Error during labeling and captioning: {e}")
+        raise
+
+    logger.info("Finished processing image with YOLO and captioning.")
+
+    # Convert the image to base64 string
     image = Image.open(io.BytesIO(base64.b64decode(dino_labled_img)))
-    print("finish processing")
     parsed_content_list_str = "\n".join(parsed_content_list)
 
-    # Encode image to base64
     buffered = io.BytesIO()
     image.save(buffered, format="PNG")
     img_str = base64.b64encode(buffered.getvalue()).decode("utf-8")
 
     return ProcessResponse(
         image=img_str,
-        parsed_content_list=str(parsed_content_list_str),
+        parsed_content_list=parsed_content_list_str,
         label_coordinates=str(label_coordinates),
     )
 
-
 @app.post("/process_image", response_model=ProcessResponse)
 async def process_image(
     image_file: UploadFile = File(...),
@@ -125,8 +128,26 @@ async def process_image(
     try:
         contents = await image_file.read()
         image_input = Image.open(io.BytesIO(contents)).convert("RGB")
+        
+        logger.info(f"Processing image: {image_file.filename}")
+        logger.info(f"Image size: {image_input.size}")
+        
+        # Debugging the input image
+        if not image_input:
+            raise ValueError("Image input is empty or invalid.")
+        
+        response = process(image_input, box_threshold, iou_threshold)
+        
+        # Ensure the response contains an image
+        if not response.image:
+            raise ValueError("Empty image in response")
+        
+        logger.info("Processing complete, returning response.")
+        return response
+        
     except Exception as e:
-        raise HTTPException(status_code=400, detail="Invalid image file")
+        logger.error(f"Error processing image: {e}")
+        import traceback
+        traceback.print_exc()
+        raise HTTPException(status_code=500, detail=str(e))
 
-    response = process(image_input, box_threshold, iou_threshold)
-    return response