# Based on https://github.com/christophschuhmann/improved-aesthetic-predictor/blob/fe88a163f4661b4ddabba0751ff645e2e620746e/simple_inference.py import os import torch import torch.nn as nn from transformers import CLIPModel, CLIPProcessor from flow_grpo.reward_ckpt_path import CKPT_PATH import numpy as np from PIL import Image class MLP(nn.Module): def __init__(self): super().__init__() self.layers = nn.Sequential( nn.Linear(768, 1024), nn.Dropout(0.2), nn.Linear(1024, 128), nn.Dropout(0.2), nn.Linear(128, 64), nn.Dropout(0.1), nn.Linear(64, 16), nn.Linear(16, 1), ) @torch.no_grad() def forward(self, embed): return self.layers(embed) class AestheticScorer(torch.nn.Module): def __init__(self, dtype, device): super().__init__() self.clip = CLIPModel.from_pretrained("openai/clip-vit-large-patch14").to(device) self.processor = CLIPProcessor.from_pretrained("openai/clip-vit-large-patch14") self.mlp = MLP().to(device) state_dict = torch.load(os.path.join(CKPT_PATH, "sac+logos+ava1-l14-linearMSE.pth"), map_location="cpu") self.mlp.load_state_dict(state_dict) self.dtype = dtype self.device = device self.eval() @torch.no_grad() def __call__(self, images): inputs = self.processor(images=images, return_tensors="pt") inputs = {k: v.to(self.dtype).to(self.device) for k, v in inputs.items()} embed = self.clip.get_image_features(**inputs) # normalize embedding embed = embed / torch.linalg.vector_norm(embed, dim=-1, keepdim=True) return self.mlp(embed).squeeze(1) # Usage example def main(): scorer = AestheticScorer(device="cuda", dtype=torch.float32) images = [ "test_cases/nasa.jpg", ] pil_images = np.stack([np.array(Image.open(img)) for img in images]) images = pil_images.transpose(0, 3, 1, 2) # NHWC -> NCHW images = torch.tensor(images, dtype=torch.uint8) print(scorer(images)) if __name__ == "__main__": main()