File size: 4,525 Bytes
133733c
 
 
 
 
 
 
 
9f9625c
133733c
9f9625c
133733c
9f9625c
eeac153
 
133733c
9f9625c
133733c
9f9625c
eeac153
9f9625c
 
 
133733c
9f9625c
133733c
9f9625c
133733c
8fabb44
133733c
9f9625c
133733c
9f9625c
 
 
133733c
0d299d4
133733c
9f9625c
133733c
9f9625c
133733c
9f9625c
133733c
9f9625c
133733c
 
9f9625c
133733c
 
 
 
 
9f9625c
133733c
 
 
 
 
 
9f9625c
 
133733c
 
 
 
 
 
 
 
9f9625c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
133733c
 
9f9625c
133733c
 
 
 
 
 
 
 
 
 
 
9f9625c
133733c
 
 
 
 
 
 
 
 
 
9f9625c
133733c
9f9625c
133733c
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
from fastapi import FastAPI, File, UploadFile, HTTPException
from fastapi.responses import JSONResponse
from pydantic import BaseModel
import base64
import io
import os
from PIL import Image
import torch
import numpy as np
from ultralytics import YOLO
from transformers import AutoProcessor, AutoModelForCausalLM

# Ensure directories exist
if not os.path.exists("weights/icon_detect"):
    os.makedirs("weights/icon_detect")

# Model loading with error handling
try:
    # Load YOLO model
    yolo_model = YOLO("weights/icon_detect/best.pt").to("cuda")
except Exception as e:
    print(f"Error loading YOLO model: {e}")
    yolo_model = YOLO("weights/icon_detect/best.pt")  # Load on CPU if CUDA fails

# Load Caption Model (Florence and OmniParser)
try:
    processor = AutoProcessor.from_pretrained("microsoft/Florence-2-base", trust_remote_code=True)
    model = AutoModelForCausalLM.from_pretrained(
        "microsoft/OmniParser",
        torch_dtype=torch.float16,
        trust_remote_code=True
    ).to("cuda")
except Exception as e:
    print(f"Error loading caption model: {e}")
    processor = AutoProcessor.from_pretrained("microsoft/Florence-2-base", trust_remote_code=True)
    model = AutoModelForCausalLM.from_pretrained(
        "microsoft/OmniParser",
        torch_dtype=torch.float16,
        trust_remote_code=True
    )

caption_model_processor = {"processor": processor, "model": model}
print("Finished loading models!")

# FastAPI app initialization
app = FastAPI()

# Pydantic response model
class ProcessResponse(BaseModel):
    image: str  # Base64 encoded image
    parsed_content_list: str
    label_coordinates: str

# Function to process the image, apply YOLO, and generate captions
def process(
    image_input: Image.Image, box_threshold: float, iou_threshold: float
) -> ProcessResponse:
    image_save_path = "imgs/saved_image_demo.png"
    image_input.save(image_save_path)
    image = Image.open(image_save_path)
    
    # Ratio for bounding box scaling
    box_overlay_ratio = image.size[0] / 3200
    draw_bbox_config = {
        "text_scale": 0.8 * box_overlay_ratio,
        "text_thickness": max(int(2 * box_overlay_ratio), 1),
        "text_padding": max(int(3 * box_overlay_ratio), 1),
        "thickness": max(int(3 * box_overlay_ratio), 1),
    }

    # OCR Box Detection and Filtering (using EasyOCR and PaddleOCR)
    try:
        ocr_bbox_rslt, is_goal_filtered = check_ocr_box(
            image_save_path,
            display_img=False,
            output_bb_format="xyxy",
            goal_filtering=None,
            easyocr_args={"paragraph": False, "text_threshold": 0.9},
            use_paddleocr=True,
        )
        text, ocr_bbox = ocr_bbox_rslt
    except Exception as e:
        raise HTTPException(status_code=500, detail=f"OCR processing failed: {e}")
    
    # YOLO and Caption Model Inference
    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:
        raise HTTPException(status_code=500, detail=f"YOLO or caption model inference failed: {e}")
    
    # Convert processed image to base64
    image = Image.open(io.BytesIO(base64.b64decode(dino_labled_img)))
    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),
        label_coordinates=str(label_coordinates),
    )

# FastAPI route to process uploaded image
@app.post("/process_image", response_model=ProcessResponse)
async def process_image(
    image_file: UploadFile = File(...),
    box_threshold: float = 0.05,
    iou_threshold: float = 0.1,
):
    try:
        contents = await image_file.read()
        image_input = Image.open(io.BytesIO(contents)).convert("RGB")
    except Exception as e:
        raise HTTPException(status_code=400, detail=f"Invalid image file: {e}")

    # Process the image
    response = process(image_input, box_threshold, iou_threshold)
    return response