app.py

import gradio as gr
import numpy as np
from PIL import Image, ImageDraw, ImageFont
import cv2
from nudenet import NudeDetector

# --- Konstanten ---
DETECTION_MAX_DIM = 768
PIXELS_PER_CM_ESTIMATE = 15
MIN_CONFIDENCE = 0.45

# --- NudeNet Detector ---
detector = NudeDetector(inference_resolution=640)

# --- Hilfsfunktionen ---
def resize_for_detection(img_pil, max_dim):
    if max(img_pil.width, img_pil.height) <= max_dim:
        return img_pil, 1.0
    ratio = max_dim / max(img_pil.width, img_pil.height)
    new_size = (int(img_pil.width * ratio), int(img_pil.height * ratio))
    resized = img_pil.resize(new_size, Image.Resampling.LANCZOS)
    scale = 1 / ratio
    return resized, scale

def describe_breast_precise(crop_pil):
    w,h = crop_pil.size
    if w*h == 0:
        return "Fehler: leeres Crop"
    gray = cv2.cvtColor(np.array(crop_pil), cv2.COLOR_RGB2GRAY)
    _, thresh = cv2.threshold(gray, 100, 255, cv2.THRESH_BINARY + cv2.THRESH_OTSU)
    contours,_ = cv2.findContours(thresh, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
    nipple_detected = any(
        40 < cv2.contourArea(c) < (w*h/4) and (p:=cv2.arcLength(c,True))>0 and
        (4*np.pi*cv2.contourArea(c)/(p*p))>0.55
        for c in contours
    )
    ratio = w/h
    shape = "Breit" if ratio>1.15 else "Hoch" if ratio<0.85 else "Rund"
    size = "klein" if w*h<28000 else "mittel" if w*h<75000 else "groß" if w*h<140000 else "sehr groß"
    w_cm = round(w/PIXELS_PER_CM_ESTIMATE,1)
    h_cm = round(h/PIXELS_PER_CM_ESTIMATE,1)
    return f"Brust: {shape}, {size}, Nippel: {'Ja' if nipple_detected else 'Nein'}, {w_cm}x{h_cm}cm"

def describe_vagina_precise(crop_pil):
    w,h = crop_pil.size
    if w*h == 0:
        return "Fehler: leeres Crop"
    gray = cv2.cvtColor(np.array(crop_pil), cv2.COLOR_RGB2GRAY)
    hair_ratio = np.sum(cv2.inRange(gray, 35, 145) > 0) / (w*h)  # <--- korrigiert
    shaved = "rasiert" if hair_ratio < 0.04 else "minimal" if hair_ratio < 0.13 else "Brazilian" if hair_ratio < 0.36 else "behaart"
    ratio = w/h
    area = w*h
    if area < 18000:
        form_desc = "Innie"
    elif area > 65000 and ratio > 1.45:
        form_desc = "Outie (Puff)"
    elif ratio > 1.45:
        form_desc = "Outie"
    else:
        form_desc = "Innie/Outie"
    size = "winzig" if area<18000 else "klein" if area<38000 else "mittel" if area<65000 else "groß"
    w_cm = round(w/PIXELS_PER_CM_ESTIMATE,1)
    h_cm = round(h/PIXELS_PER_CM_ESTIMATE,1)
    return f"Vagina: {form_desc}, {size}, {shaved}, {w_cm}x{h_cm}cm"

# --- Bildverarbeitung ---
def process_image(image):
    try:
        original_pil = Image.fromarray(image).convert("RGB") if isinstance(image,np.ndarray) else image.convert("RGB")
        detection_pil, scale = resize_for_detection(original_pil, DETECTION_MAX_DIM)
        detections = detector.detect(np.array(detection_pil))
        draw = ImageDraw.Draw(original_pil)
        font = ImageFont.load_default()
        results_text = []

        for det in detections:
            label = det["class"]
            score = det.get("score",0)
            if score < MIN_CONFIDENCE:
                continue
            if label not in ["FEMALE_BREAST_EXPOSED","FEMALE_GENITALIA_EXPOSED"]:
                continue
            x,y,w,h = [int(v*scale) for v in det["box"]]
            crop_pil = original_pil.crop((x,y,x+w,y+h))
            if label=="FEMALE_BREAST_EXPOSED":
                desc = describe_breast_precise(crop_pil)
                color = (255,46,130)
            else:
                desc = describe_vagina_precise(crop_pil)
                color = (138,43,226)
            draw.rectangle([x,y,x+w,y+h],outline=color,width=4)
            text_pos = (x,y-15 if y>15 else y+h)
            draw.text(text_pos,desc,fill=color,font=font)
            results_text.append(desc)

        if not results_text:
            draw.text((10,10),"Keine relevanten Bereiche erkannt.",fill=(255,0,0),font=font)

        return np.array(original_pil)

    except Exception as e:
        print(f"Fehler: {e}")
        return None

# --- Gradio App ---
css = """
body { background: #0f0f1a; color: #e0e0ff; }
.gradio-container { max-width: 900px !important; margin: auto; }
h1 { color: #ff2e82; text-align: center; }
"""

with gr.Blocks(css=css) as demo:
    gr.Markdown("# 👙 Automatischer Nackt-Analyzer")
    gr.Markdown("Lade ein Bild hoch und erhalte direkt das analysierte Bild mit Annotationen.")
    input_image = gr.Image(type="numpy", label="Bild hochladen")
    output_image = gr.Image(label="Analyse-Ergebnis")
    input_image.change(fn=process_image, inputs=input_image, outputs=output_image)

demo.launch()