app.py

import gradio as gr
import numpy as np
from PIL import Image
from ultralytics import YOLO
import cv2
from pathlib import Path

path = Path(__file__).parent

# Model configurations
MODEL_CONFIGS = {
    "Dry Season Form": {
        "path": path/"models/DSF_Mleda_250.pt",
        "labels": [
            "F1", "F2", "F3", "F4", "F5", "F6", "F7", "F8", "F9", "F10", "F11", "F12", "F13", "F14", "F15", "F16",
            "H1", "H2", "H3", "H4", "H5", "H6", "H7", "H8", "H9", "H10", "H11", "H12", "H13", "H14", "H15", "H16", "H17",
            "fs1", "fs2", "fs3", "fs4", "fs5",
            "hs1", "hs2", "hs3", "hs4", "hs5", "hs6",
            "sc1", "sc2",
            "sex", "right", "left", "grey", "black", "white"
        ],
        "imgsz": 1280
    },
    "Wet Season Form": {
        "path": path/"models/WSF_Mleda_200.pt",
        "labels": [
            'F12', 'F14', 'fs1', 'fs2', 'fs3', 'fs4', 'fs5',
            'H12', 'H14', 'hs1', 'hs2', 'hs3', 'hs4', 'hs5', 'hs6',
            'white', 'black', 'grey', 'sex', 'blue', 'green', 'red', 'sc1', 'sc2'
        ],
        "imgsz": 1280
    },
    "All Season Form": {
        "path": path/"models/DSF_WSF_Mleda_450.pt",
        "labels": [
            "F1", "F2", "F3", "F4", "F5", "F6", "F7", "F8", "F9", "F10", "F11", "F12", "F13", "F14", "F15", "F16",
            "H1", "H2", "H3", "H4", "H5", "H6", "H7", "H8", "H9", "H10", "H11", "H12", "H13", "H14", "H15", "H16", "H17",
            "fs1", "fs2", "fs3", "fs4", "fs5",
            "hs1", "hs2", "hs3", "hs4", "hs5", "hs6",
            "sc1", "sc2",
            "sex", "right", "left", "grey", "black", "white"
        ],
        "imgsz": 1280
    }
}

def hex_to_bgr(hex_color: str) -> tuple:
    """Convert #RRGGBB hex color to BGR tuple."""
    hex_color = hex_color.lstrip("#")
    if len(hex_color) != 6:
        return (0, 255, 0)  # Default to green if invalid
    r = int(hex_color[0:2], 16)
    g = int(hex_color[2:4], 16)
    b = int(hex_color[4:6], 16)
    return (b, g, r)

def load_model(path):
    """Load YOLO model from the given path."""
    return YOLO(path)

def draw_detections(image: np.ndarray, results, labels, keypoint_threshold: float, show_labels: bool, point_size: int, point_color: str, label_size: float) -> np.ndarray:
    img = image.copy()
    color_bgr = hex_to_bgr(point_color)

    for result in results:
        boxes = result.boxes.xywh.cpu().numpy()
        cls_ids = result.boxes.cls.int().cpu().numpy()
        confs = result.boxes.conf.cpu().numpy()
        kpts_all = result.keypoints.data.cpu().numpy()

        for (x_c, y_c, w, h), cls_id, conf, kpts in zip(boxes, cls_ids, confs, kpts_all):
            x1 = int(x_c - w/2); y1 = int(y_c - h/2)
            x2 = int(x_c + w/2); y2 = int(y_c + h/2)

            cv2.rectangle(img, (x1, y1), (x2, y2), color=(255,255,0), thickness=2)

            class_name = result.names[int(cls_id)]
            text = f"{class_name} {conf:.2f}"
            (tw, th), _ = cv2.getTextSize(text, cv2.FONT_HERSHEY_SIMPLEX, 0.6, 1)
            cv2.rectangle(img, (x1, y1 - th - 4), (x1 + tw, y1), (255,255,0), cv2.FILLED)
            cv2.putText(img, text, (x1, y1 - 4), cv2.FONT_HERSHEY_SIMPLEX, 0.6, (0,0,0), 1, cv2.LINE_AA)

            for i, (x, y, v) in enumerate(kpts):
                if v > keypoint_threshold and i < len(labels):
                    xi, yi = int(x), int(y)
                    cv2.circle(img, (xi, yi), int(point_size), color_bgr, -1, cv2.LINE_AA)
                    if show_labels:
                        cv2.putText(img, labels[i], (xi + 3, yi + 3), cv2.FONT_HERSHEY_SIMPLEX, label_size, (255,0,0), 2, cv2.LINE_AA)
    return img


def predict_and_annotate(input_image: Image.Image, conf_threshold: float, keypoint_threshold: float, model_choice: str, show_labels: bool, point_size: int, point_color: str, label_size: float):
    config = MODEL_CONFIGS[model_choice]
    model = load_model(config["path"])
    labels = config["labels"]
    imgsz = config["imgsz"]

    img_rgb = np.array(input_image.convert("RGB"))
    img_bgr = cv2.cvtColor(img_rgb, cv2.COLOR_RGB2BGR)

    results = model(img_bgr, conf=conf_threshold, imgsz=imgsz)
    annotated_bgr = draw_detections(img_bgr, results, labels, keypoint_threshold, show_labels, point_size, point_color, label_size)
    annotated_rgb = cv2.cvtColor(annotated_bgr, cv2.COLOR_BGR2RGB)
    return Image.fromarray(annotated_rgb)


# Gradio Interface
app = gr.Interface(
    fn=predict_and_annotate,
    inputs=[
        gr.Image(type="pil", label="Upload Image"),
        gr.Slider(minimum=0.0, maximum=1.0, value=0.25, step=0.01, label="Confidence Threshold"),
        gr.Slider(minimum=0.0, maximum=1.0, value=0.5, step=0.01, label="Keypoint Visibility Threshold"),
        gr.Radio(choices=list(MODEL_CONFIGS.keys()), label="Select Model", value="Dry Season Form"),
        gr.Checkbox(label="Show Keypoint Labels", value=True),
        gr.Slider(minimum=1, maximum=20, value=8, step=0.1, label="Keypoint Size"),
        gr.ColorPicker(label="Keypoint Color", value="#00FF00"),
        gr.Slider(minimum=0.3, maximum=3.0, value=1.0, step=0.1, label="Keypoint Label Font Size")
    ],
    outputs=gr.Image(type="pil", label="Detection Result", format="png"),
    title="🦋 Melanitis leda Landmark Identification",
    description="Upload an image and select the model. Customize detection and keypoint display settings.",
    flagging_mode="never"
)


if __name__ == "__main__":
    app.launch()