app.py

import gradio as gr
import tempfile
import os
import cv2
from sahi import AutoDetectionModel
from sahi.predict import get_sliced_prediction
from PIL import Image
import numpy as np

# Inisialisasi model deteksi menggunakan SAHI
model_path = "best.pt"  # Ganti dengan path model YOLO lokal Anda
confidence_threshold = 0.6  # Threshold kepercayaan
sahi_device = 'cuda'  # Ganti dengan 'cpu' jika tidak menggunakan GPU

# Memuat model YOLO menggunakan SAHI
sahi_model = AutoDetectionModel.from_pretrained(
    model_type="yolov11",  # Tipe model YOLO, sesuaikan jika model YOLO yang digunakan berbeda
    model_path=model_path,
    confidence_threshold=confidence_threshold,
    device=sahi_device
)

# Fungsi untuk deteksi objek menggunakan SAHI
def detect_objects(image):
    # Simpan gambar yang diupload sebagai file sementara
    with tempfile.NamedTemporaryFile(delete=False, suffix=".jpg") as temp_file:
        image.save(temp_file, format="JPEG")
        temp_file_path = temp_file.name

    try:
        # Lakukan prediksi pada gambar menggunakan SAHI
        results = get_sliced_prediction(
            image=image,
            detection_model=sahi_model,
            slice_height=512,  # Ukuran potongan gambar (bisa disesuaikan)
            slice_width=512,
            overlap_height_ratio=0.2,
            overlap_width_ratio=0.2
        )

        # Menghitung jumlah objek per kelas
        class_count = {}
        total_count = 0  # Menyimpan total jumlah objek

        # Menggambar bounding boxes pada gambar
        output_image = np.array(image)  # Convert PIL Image to numpy array for OpenCV processing

        for prediction in results.object_prediction_list:
            bbox = prediction.bbox
            class_name = prediction.category.name  # Nama kelas objek
            confidence = prediction.score.value  # Skor prediksi
            
            # Hanya gambar bounding box jika skor kepercayaan lebih besar dari threshold
            if confidence >= confidence_threshold:
                # Gambar bounding box
                cv2.rectangle(output_image, 
                              (int(bbox.minx), int(bbox.miny)), 
                              (int(bbox.maxx), int(bbox.maxy)), 
                              (0, 255, 0), 2)  # Gambar kotak hijau

                # Gambar label dan skor
                cv2.putText(output_image, 
                            f"{class_name} {confidence:.2f}", 
                            (int(bbox.minx), int(bbox.miny) - 10),
                            cv2.FONT_HERSHEY_SIMPLEX, 0.9, 
                            (0, 255, 0), 2)

                # Hitung jumlah objek per kelas
                class_count[class_name] = class_count.get(class_name, 0) + 1
                total_count += 1  # Menambah jumlah objek

        # Menyusun output berupa string hasil perhitungan
        result_text = "Detected Objects:\n\n"
        for class_name, count in class_count.items():
            result_text += f"{class_name}: {count}\n"
        result_text += f"\nTotal Objects: {total_count}"

        # Convert output_image (numpy array) back to PIL Image to save
        output_image_pil = Image.fromarray(output_image)
        output_image_path = "/tmp/prediction.jpg"
        output_image_pil.save(output_image_path)  # Menyimpan gambar dengan prediksi

    except Exception as err:
        # Menangani kesalahan lain
        result_text = f"An error occurred: {err}"
        output_image_path = temp_file_path  # Kembalikan gambar asli jika terjadi error

    # Hapus file sementara setelah prediksi
    os.remove(temp_file_path)

    return output_image_path, result_text

# Membuat antarmuka Gradio dengan tata letak fleksibel
with gr.Blocks() as iface:
    with gr.Row():
        with gr.Column():
            input_image = gr.Image(type="pil", label="Input Image")
        with gr.Column():
            output_image = gr.Image(label="Detect Object")
        with gr.Column():
            output_text = gr.Textbox(label="Counting Object")

    # Tombol untuk memproses input
    detect_button = gr.Button("Detect")

    # Hubungkan tombol dengan fungsi deteksi
    detect_button.click(
        fn=detect_objects, 
        inputs=input_image, 
        outputs=[output_image, output_text]
    )

# Menjalankan antarmuka
iface.launch()