81 lines
3.2 KiB
Python
81 lines
3.2 KiB
Python
import json
|
|
from pydantic import BaseModel, Field
|
|
from APIs.aichat import generate
|
|
import shared.config as config
|
|
from PIL import Image, ImageEnhance
|
|
|
|
class UtensilMaterialCheck(BaseModel):
|
|
object_name: str = Field(
|
|
description="Name of utensil or container seen (e.g., spoon, fork, tray)."
|
|
)
|
|
is_metal: bool = Field(
|
|
description="True ONLY if the object is made of metal, stainless steel, or aluminum."
|
|
)
|
|
|
|
class DishSafetyResult(BaseModel):
|
|
utensils_and_containers: list[UtensilMaterialCheck] = Field(
|
|
description="List each visible non-food object and check if it is made of metal."
|
|
)
|
|
material_analysis: str = Field(
|
|
description="Short description of the materials present."
|
|
)
|
|
detected_hazards: list[str] = Field(
|
|
description="List objects from utensils_and_containers where is_metal is True."
|
|
)
|
|
is_safe: bool = Field(
|
|
description="True if detected_hazards is completely empty []. False if ANY metal is found."
|
|
)
|
|
warning_message: str = Field(
|
|
description="If is_safe is False, set to 'REMOVE METAL UTENSIL OR FOIL BEFORE MICROWAVING'. Otherwise empty ''."
|
|
)
|
|
|
|
def preprocess_dish_image(image_path: str) -> str:
|
|
"""Brightens dark areas and sharpens food texture for small vision models."""
|
|
img = Image.open(image_path)
|
|
|
|
# 1. Boost brightness slightly
|
|
enhancer = ImageEnhance.Brightness(img)
|
|
img = enhancer.enhance(1.4)
|
|
|
|
# 2. Boost contrast to separate food shapes from dark shadows
|
|
enhancer = ImageEnhance.Contrast(img)
|
|
img = enhancer.enhance(1.3)
|
|
|
|
processed_path = "/tmp/processed_dish.jpg"
|
|
img.save(processed_path, quality=85)
|
|
return processed_path
|
|
|
|
def check_dish_safety(image_path: str) -> dict:
|
|
prompt = (
|
|
"Analyze this microwave dish image strictly for EQUIPMENT DAMAGE HAZARDS (Metal, Steel, Aluminum Foil).\n\n"
|
|
"GOAL:\n"
|
|
"Identify if any metallic item (metal spoon, fork, knife, aluminum foil, wire) is inside the dish.\n\n"
|
|
"RULES:\n"
|
|
"1. Examine all utensils carefully, even in dim lighting or dark spots.\n"
|
|
"2. Spoons, forks, and knives are often STAINLESS STEEL / METAL. If a spoon or fork is visible, mark is_metal = True unless it is clearly bright colored plastic.\n"
|
|
"3. Ignore chemical or health concerns (plastic cancer risks are IRRELEVANT). Focus ONLY on spark/fire hazards (metal, foil).\n"
|
|
"4. If ANY utensil is metal: set is_metal=True, add it to detected_hazards, and set is_safe=False."
|
|
)
|
|
|
|
response_raw = generate(
|
|
prompt=prompt,
|
|
images=[image_path],
|
|
output_format=DishSafetyResult,
|
|
should_think=False,
|
|
)
|
|
print(f"[Debug] Raw response from AI generator: {response_raw}")
|
|
|
|
if isinstance(response_raw, str):
|
|
try:
|
|
validated_result = DishSafetyResult.model_validate_json(response_raw)
|
|
return validated_result.model_dump()
|
|
except Exception:
|
|
return json.loads(response_raw)
|
|
|
|
if isinstance(response_raw, DishSafetyResult):
|
|
return response_raw.model_dump()
|
|
|
|
if isinstance(response_raw, dict):
|
|
return response_raw
|
|
|
|
raise ValueError(f"Unexpected response type from AI generator: {type(response_raw)}") |