DL-Art-School/codes/trainer/eval/sr_diffusion_fid.py

63 lines
2.8 KiB
Python

import os
import torch
import os.path as osp
import torchvision
from torch.nn.functional import interpolate
from tqdm import tqdm
import trainer.eval.evaluator as evaluator
from pytorch_fid import fid_score
from data import create_dataset
from torch.utils.data import DataLoader, DistributedSampler, SequentialSampler
from trainer.injectors.gaussian_diffusion_injector import GaussianDiffusionInferenceInjector
from utils.util import opt_get
# Performs a FID evaluation on a diffusion network
class SrDiffusionFidEvaluator(evaluator.Evaluator):
def __init__(self, model, opt_eval, env):
super().__init__(model, opt_eval, env)
self.batch_sz = opt_eval['batch_size']
self.fid_batch_size = opt_get(opt_eval, ['fid_batch_size'], 64)
assert self.batch_sz is not None
self.dataset = create_dataset(opt_eval['dataset'])
if torch.distributed.is_available() and torch.distributed.is_initialized():
self.sampler = DistributedSampler(self.dataset, shuffle=False, drop_last=True)
else:
self.sampler = SequentialSampler(self.dataset)
self.fid_real_samples = opt_eval['dataset']['paths'] # This is assumed to exist for the given dataset.
assert isinstance(self.fid_real_samples, str)
self.gd = GaussianDiffusionInferenceInjector(opt_eval['diffusion_params'], env)
self.out_key = opt_eval['diffusion_params']['out']
def perform_eval(self):
# Attempt to make the dataset deterministic.
self.dataset.reset_random()
dataloader = DataLoader(self.dataset, self.batch_sz, sampler=self.sampler, num_workers=0)
fid_fake_path = osp.join(self.env['base_path'], "..", "fid", str(self.env["step"]))
os.makedirs(fid_fake_path, exist_ok=True)
counter = 0
for batch in tqdm(dataloader):
batch = {k: v.to(self.env['device']) if isinstance(v, torch.Tensor) else v for k, v in batch.items()}
gen = self.gd(batch)[self.out_key]
# All gather if we're in distributed mode.
if torch.distributed.is_available() and torch.distributed.is_initialized():
gather_list = [torch.zeros_like(gen) for _ in range(torch.distributed.get_world_size())]
torch.distributed.all_gather(gather_list, gen)
gen = torch.cat(gather_list, dim=0)
if self.env['rank'] <= 0:
for g in gen:
torchvision.utils.save_image(g, osp.join(fid_fake_path, f"{counter}.png"))
counter += 1
if self.env['rank'] <= 0:
return {"fid": fid_score.calculate_fid_given_paths([self.fid_real_samples, fid_fake_path], self.fid_batch_size,
True, 2048)}
else:
return {}