""" Image Enhancer App ------------------ Aplikasi berbasis Gradio untuk meningkatkan kualitas gambar dengan berbagai teknik pemrosesan citra seperti perbaikan resolusi, penajaman, dan penghilangan noise. Dapat digunakan secara lokal maupun sebagai web-app AI image enhancement. Created by _drat | 2025 """ # Import library standar dan eksternal yang dibutuhkan import os import subprocess import spaces # Library dari HuggingFace untuk GPU management & dekorator import torch # Library untuk deep learning, cek GPU import cv2 # OpenCV untuk pengolahan gambar import uuid # Untuk generate nama file unik import gradio as gr # Untuk membuat UI web berbasis Python import numpy as np # Operasi numerik, termasuk array from PIL import Image # Untuk memproses dan menyimpan gambar # Import arsitektur model dan utilitas untuk enhancement & upscaling from basicsr.archs.srvgg_arch import SRVGGNetCompact from gfpgan.utils import GFPGANer from realesrgan.utils import RealESRGANer import warnings warnings.filterwarnings("ignore") # Fungsi untuk menjalankan command shell (misal: wget file model) def runcmd(cmd, verbose = False): process = subprocess.Popen( cmd, stdout = subprocess.PIPE, stderr = subprocess.PIPE, text = True, shell = True ) std_out, std_err = process.communicate() if verbose: print(std_out.strip(), std_err) pass # Pastikan model AI (GFPGAN dan RealESRGAN) sudah terdownload. Jika belum, otomatis download. if not os.path.exists('GFPGANv1.4.pth'): runcmd("wget https://github.com/TencentARC/GFPGAN/releases/download/v1.3.0/GFPGANv1.4.pth -P .") if not os.path.exists('realesr-general-x4v3.pth'): runcmd("wget https://github.com/xinntao/Real-ESRGAN/releases/download/v0.2.5.0/realesr-general-x4v3.pth -P .") # Inisialisasi model Real-ESRGAN untuk upscaling gambar model = SRVGGNetCompact(num_in_ch=3, num_out_ch=3, num_feat=64, num_conv=32, upscale=4, act_type='prelu') model_path = 'realesr-general-x4v3.pth' half = True if torch.cuda.is_available() else False # Pakai mode half-precision jika ada GPU (lebih cepat) upsampler = RealESRGANer( scale=4, model_path=model_path, model=model, tile=0, tile_pad=10, pre_pad=0, half=half ) # Fungsi utama untuk enhance gambar (dipanggil saat tombol di UI ditekan) @spaces.GPU(duration=15) def enhance_image( input_image: Image, # Gambar input dari user (PIL Image) scale: int, # Skala upscaling (misal: 2x, 4x) enhance_mode: str, # Mode enhance: face saja, image saja, atau keduanya ): only_face = enhance_mode == "Only Face Enhance" # Pilih mode enhancer: hanya wajah, hanya gambar, atau kombinasi if enhance_mode == "Only Face Enhance": face_enhancer = GFPGANer( model_path='GFPGANv1.4.pth', upscale=scale, arch='clean', channel_multiplier=2 ) elif enhance_mode == "Only Image Enhance": face_enhancer = None # Tidak enhance wajah, hanya upscaling gambar saja else: face_enhancer = GFPGANer( model_path='GFPGANv1.4.pth', upscale=scale, arch='clean', channel_multiplier=2, bg_upsampler=upsampler ) # Konversi gambar input ke format BGR (OpenCV) img = cv2.cvtColor(np.array(input_image), cv2.COLOR_RGB2BGR) h, w = img.shape[0:2] # Optional: perbesar gambar jika terlalu kecil (kurang dari 300px) if h < 300: img = cv2.resize(img, (w * 2, h * 2), interpolation=cv2.INTER_LANCZOS4) # Proses enhance gambar (tergantung mode) if face_enhancer is not None: # Enhance wajah (GFPGAN), output gambar hasil enhancement _, _, output = face_enhancer.enhance( img, has_aligned=False, only_center_face=only_face, paste_back=True ) else: # Hanya upscaling image (tanpa enhance wajah) output, _ = upsampler.enhance(img, outscale=scale) # --- Optional resize (tidak digunakan) # if scale != 2: # interpolation = cv2.INTER_AREA if scale < 2 else cv2.INTER_LANCZOS4 # h, w = img.shape[0:2] # output = cv2.resize(output, (int(w * scale / 2), int(h * scale / 2)), interpolation=interpolation) # Batasi hasil enhance agar tidak lebih dari 3480px (biar aman dan efisien) h, w = output.shape[0:2] max_size = 3480 if h > max_size: w = int(w * max_size / h) h = max_size if w > max_size: h = int(h * max_size / w) w = max_size # Resize hasil akhir output = cv2.resize(output, (w, h), interpolation=cv2.INTER_LANCZOS4) # Konversi kembali ke RGB & PIL Image (untuk output Gradio) enhanced_image = Image.fromarray(cv2.cvtColor(output, cv2.COLOR_BGR2RGB)) # Simpan gambar hasil enhance ke folder output dengan nama unik tmpPrefix = "/tmp/gradio/" extension = 'png' # targetDir = f"{tmpPrefix}output/" targetDir = "output/" if not os.path.exists(targetDir): os.makedirs(targetDir) enhanced_path = f"{targetDir}{uuid.uuid4()}.{extension}" enhanced_image.save(enhanced_path, quality=100) return enhanced_image, enhanced_path # Return: image untuk preview dan path file untuk download # # DEMO DI SINI # def create_demo() -> gr.Blocks: with gr.Blocks() as demo: # Feature Image dengan border subtle & shadow gr.HTML( """
Feature Image
""" ) # Judul & Tagline dengan emoji & gradient text gr.Markdown("""
📸 Image Enhancer Pro ✨
🤖 Foto blur? Jadikan jernih & tajam pakai AI!
""") # Input dan Output dalam card, dengan icon pada title with gr.Row(): with gr.Column(scale=1): with gr.Group(elem_id="input-card"): gr.Markdown("
🖼️ Gambar Asli
") input_image = gr.Image(label="", type="pil", elem_id="input-image", show_label=False) gr.Markdown("
⬆️ Upload foto format JPG/PNG, max 5MB
") with gr.Column(scale=1): with gr.Group(elem_id="output-card"): gr.Markdown("
🌈 Hasil AI Enhance
") output_image = gr.Image(label="", type="pil", interactive=False, elem_id="output-image", show_label=False) enhance_image_path = gr.File(label="⬇️ Download (.png)", interactive=False, elem_id="download-btn") gr.Markdown("""
💡 Klik untuk download hasilnya!
""") # Kontrol dengan ikon pada label with gr.Group(elem_id="control-card"): with gr.Row(): scale = gr.Slider( minimum=1, maximum=4, value=2, step=1, label="🔎 Upscale x", elem_id="scale-slider" ) enhance_mode = gr.Dropdown( label="🛠️ Mode Enhance", choices=[ "Only Face Enhance", "Only Image Enhance", "Face Enhance + Image Enhance", ], value="Face Enhance + Image Enhance", elem_id="mode-dropdown" ) g_btn = gr.Button("🚀 Enhance Sekarang!", elem_id="enhance-btn", size="lg") g_btn.click( fn=enhance_image, inputs=[input_image, scale, enhance_mode], outputs=[output_image, enhance_image_path], ) # Footer dengan ikon & sentuhan branding # Tambahkan footer di bagian bawah gr.HTML(""" """) return demo