tiny_gfpgan.py

#######################################################################################
#
# MIT License
#
# Copyright (c) [2025] [leonelhs@gmail.com]
#
# Permission is hereby granted, free of charge, to any person obtaining a copy
# of this software and associated documentation files (the "Software"), to deal
# in the Software without restriction, including without limitation the rights
# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
# copies of the Software, and to permit persons to whom the Software is
# furnished to do so, subject to the following conditions:
#
# The above copyright notice and this permission notice shall be included in all
# copies or substantial portions of the Software.
#
# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
# SOFTWARE.
#
#######################################################################################

# This file implements an API endpoint GFPGAN system.
# It provides functionality to enhances an image by upscaling it 4 times its original size .


# Source code is based on or inspired by several projects.
# For more details and proper attribution, please refer to the following resources:
#
# - [GFPGAN] - [https://github.com/TencentARC/GFPGAN]
# - [utils.py] - [https://github.com/TencentARC/GFPGAN/blob/master/gfpgan/utils.py]


import torch
from basicsr.utils import img2tensor, tensor2img
from gfpgan.archs.gfpganv1_clean_arch import GFPGANv1Clean
from huggingface_hub import hf_hub_download
from torchvision.transforms.functional import normalize

GFPGAN_REPO_ID = 'leonelhs/gfpgan'

class TinyGFPGAN:
    """
    Minimal functionalities from GFPGAN project.

    Args:
        channel_multiplier (int): Channel multiplier for large networks of StyleGAN2. Default: 2.
    """

    def __init__(self, channel_multiplier=2, device=None):

        # initialize model
        self.device = torch.device('cuda' if torch.cuda.is_available() else 'cpu') if device is None else device
        # initialize the GFP-GAN
        self.gfpgan = GFPGANv1Clean(
            out_size=512,
            channel_multiplier=channel_multiplier,
            fix_decoder=False,
            input_is_latent=True,
            different_w=True,
            sft_half=True)

        model_path = hf_hub_download(repo_id=GFPGAN_REPO_ID, filename="GFPGANv1.4.pth")
        loadnet = torch.load(model_path)
        if 'params_ema' in loadnet:
            keyname = 'params_ema'
        else:
            keyname = 'params'
        self.gfpgan.load_state_dict(loadnet[keyname], strict=True)
        self.gfpgan.eval()
        self.gfpgan = self.gfpgan.to(self.device)

    @torch.no_grad()
    def inference(self, cropped_face, weight=0.5):
        cropped_face_t = img2tensor(cropped_face / 255., bgr2rgb=True, float32=True)
        normalize(cropped_face_t, [0.5, 0.5, 0.5], [0.5, 0.5, 0.5], inplace=True)
        cropped_face_t = cropped_face_t.unsqueeze(0).to(self.device)

        try:
            output = self.gfpgan(cropped_face_t, return_rgb=False, weight=weight)[0]
            # convert to image
            restored_face = tensor2img(output.squeeze(0), rgb2bgr=True, min_max=(-1, 1))
            return restored_face.astype('uint8')
        except RuntimeError as error:
            raise ValueError(f'\tFailed inference for GFPGAN: {error}.')

    def enhance(self, cropped_faces):
        return [self.inference(face) for face in cropped_faces]