DL-Art-School/codes/process_video.py
James Betker 11155aead4 Directly use dataset keys
This has been a long time coming. Cleans up messy "GT" nomenclature and simplifies ExtensibleTraner.feed_data
2020-12-04 20:14:53 -07:00

217 lines
9.1 KiB
Python

import argparse
import logging
import os
import os.path as osp
import subprocess
import time
import torch
import torch.utils.data as data
import torchvision.transforms.functional as F
from PIL import Image
from tqdm import tqdm
from models.ExtensibleTrainer import ExtensibleTrainer
from utils import options as option
import utils.util as util
from data import create_dataloader
class FfmpegBackedVideoDataset(data.Dataset):
'''Pulls frames from a video one at a time using FFMPEG.'''
def __init__(self, opt, working_dir):
super(FfmpegBackedVideoDataset, self).__init__()
self.opt = opt
self.video = self.opt['video_file']
self.working_dir = working_dir
self.frame_rate = self.opt['frame_rate']
self.start_at = self.opt['start_at_seconds']
self.end_at = self.opt['end_at_seconds']
self.force_multiple = self.opt['force_multiple']
self.frame_count = (self.end_at - self.start_at) * self.frame_rate
# The number of (original) video frames that will be stored on the filesystem at a time.
self.max_working_files = 20
self.data_type = self.opt['data_type']
self.vertical_splits = self.opt['vertical_splits'] if 'vertical_splits' in opt.keys() else 1
def get_time_for_it(self, it):
secs = it / self.frame_rate + self.start_at
mins = int(secs / 60)
hours = int(mins / 60)
secs = secs - (mins * 60) - (hours * 3600)
mins = mins % 60
return '%02d:%02d:%06.3f' % (hours, mins, secs)
def __getitem__(self, index):
if self.vertical_splits > 0:
actual_index = int(index / self.vertical_splits)
else:
actual_index = index
# Extract the frame. Command template: `ffmpeg -ss 17:00.0323 -i <video file>.mp4 -vframes 1 destination.png`
working_file_name = osp.join(self.working_dir, "working_%d.png" % (actual_index % self.max_working_files,))
vid_time = self.get_time_for_it(actual_index)
ffmpeg_args = ['ffmpeg', '-y', '-ss', vid_time, '-i', self.video, '-vframes', '1', working_file_name]
process = subprocess.Popen(ffmpeg_args, stderr=subprocess.DEVNULL, stdout=subprocess.DEVNULL)
process.wait()
# get LQ image
LQ_path = working_file_name
img_LQ = Image.open(LQ_path)
split_index = (index % self.vertical_splits)
if self.vertical_splits > 0:
w, h = img_LQ.size
w_per_split = int(w / self.vertical_splits)
left = w_per_split * split_index
img_LQ = F.crop(img_LQ, 0, left, h, w_per_split)
img_LQ = F.to_tensor(img_LQ)
mask = torch.ones(1, img_LQ.shape[1], img_LQ.shape[2])
ref = torch.cat([img_LQ, mask], dim=0)
if self.force_multiple > 1:
assert self.vertical_splits <= 1 # This is not compatible with vertical splits for now.
c, h, w = img_LQ.shape
h_, w_ = h, w
height_removed = h % self.force_multiple
width_removed = w % self.force_multiple
if height_removed != 0:
h_ = self.force_multiple * ((h // self.force_multiple) + 1)
if width_removed != 0:
w_ = self.force_multiple * ((w // self.force_multiple) + 1)
lq_template = torch.zeros(c,h_,w_)
lq_template[:,:h,:w] = img_LQ
ref_template = torch.zeros(c,h_,w_)
ref_template[:,:h,:w] = img_LQ
img_LQ = lq_template
ref = ref_template
return {'lq': img_LQ, 'lq_fullsize_ref': ref,
'lq_center': torch.tensor([img_LQ.shape[1] // 2, img_LQ.shape[2] // 2], dtype=torch.long) }
def __len__(self):
return self.frame_count * self.vertical_splits
def merge_images(files, output_path):
"""Merges several image files together across the vertical axis
"""
images = [Image.open(f) for f in files]
w, h = images[0].size
result_width = w * len(images)
result_height = h
result = Image.new('RGB', (result_width, result_height))
for i in range(len(images)):
result.paste(im=images[i], box=(i * w, 0))
result.save(output_path)
if __name__ == "__main__":
#### options
torch.backends.cudnn.benchmark = True
want_just_images = True
parser = argparse.ArgumentParser()
parser.add_argument('-opt', type=str, help='Path to options YMAL file.', default='../options/use_video_upsample.yml')
opt = option.parse(parser.parse_args().opt, is_train=False)
opt = option.dict_to_nonedict(opt)
util.mkdirs(
(path for key, path in opt['path'].items()
if not key == 'experiments_root' and 'pretrain_model' not in key and 'resume' not in key))
util.setup_logger('base', opt['path']['log'], 'test_' + opt['name'], level=logging.INFO,
screen=True, tofile=True)
logger = logging.getLogger('base')
logger.info(option.dict2str(opt))
util.loaded_options = opt
#### Create test dataset and dataloader
test_loaders = []
test_set = FfmpegBackedVideoDataset(opt['dataset'], opt['path']['results_root'])
test_loader = create_dataloader(test_set, opt['dataset'])
logger.info('Number of test images in [{:s}]: {:d}'.format(opt['dataset']['name'], len(test_set)))
test_loaders.append(test_loader)
model = ExtensibleTrainer(opt)
test_set_name = test_loader.dataset.opt['name']
logger.info('\nTesting [{:s}]...'.format(test_set_name))
test_start_time = time.time()
dataset_dir = osp.join(opt['path']['results_root'], test_set_name)
util.mkdir(dataset_dir)
frame_counter = 0
frames_per_vid = opt['frames_per_mini_vid']
minivid_crf = opt['minivid_crf']
vid_output = opt['mini_vid_output_folder'] if 'mini_vid_output_folder' in opt.keys() else dataset_dir
vid_counter = opt['minivid_start_no'] if 'minivid_start_no' in opt.keys() else 0
img_index = opt['generator_img_index']
recurrent_mode = opt['recurrent_mode']
if recurrent_mode:
assert opt['dataset']['batch_size'] == 1 # Can only do 1 frame at a time in recurrent mode, by definition.
scale = opt['scale']
first_frame = True
ffmpeg_proc = None
tq = tqdm(test_loader)
for data in tq:
need_GT = False if test_loader.dataset.opt['dataroot_GT'] is None else True
if recurrent_mode and first_frame:
b, c, h, w = data['lq'].shape
recurrent_entry = torch.zeros((b,c,h*scale,w*scale), device=data['lq'].device)
# Optionally swap out the 'generator' for the first frame to create a better image that the recurrent generator works off of.
if 'recurrent_hr_generator' in opt.keys():
recurrent_gen = model.env['generators']['generator']
model.env['generators']['generator'] = model.env['generators'][opt['recurrent_hr_generator']]
first_frame = False
if recurrent_mode:
data['recurrent'] = recurrent_entry
model.feed_data(data, need_GT=need_GT)
model.test()
visuals = model.get_current_visuals()['rlt']
if recurrent_mode:
recurrent_entry = visuals
visuals = visuals.cpu().float()
for i in range(visuals.shape[0]):
sr_img = util.tensor2img(visuals[i]) # uint8
# save images
save_img_path = osp.join(dataset_dir, '%08d.png' % (frame_counter,))
util.save_img(sr_img, save_img_path)
frame_counter += 1
if frame_counter % frames_per_vid == 0:
if ffmpeg_proc is not None:
print("Waiting for last encode..")
ffmpeg_proc.wait()
print("Encoding minivid %d.." % (vid_counter,))
# Perform stitching.
num_splits = opt['dataset']['vertical_splits'] if 'vertical_splits' in opt['dataset'].keys() else 1
if num_splits > 1:
procs = []
src_imgs_path = osp.join(dataset_dir, "joined")
os.makedirs(src_imgs_path, exist_ok=True)
for i in range(int(frames_per_vid / num_splits)):
to_join = [osp.join(dataset_dir, "%08d.png" % (j,)) for j in range(i * num_splits, i * num_splits + num_splits)]
merge_images(to_join, osp.join(src_imgs_path, "%08d.png" % (i,)))
else:
src_imgs_path = dataset_dir
# Encoding command line:
# ffmpeg -framerate 30 -i %08d.png -c:v libx265 -crf 12 -preset slow -pix_fmt yuv444p test.mkv
cmd = ['ffmpeg', '-y', '-framerate', str(opt['dataset']['frame_rate']), '-f', 'image2', '-i', osp.join(src_imgs_path, "%08d.png"),
'-c:v', 'libx265', '-crf', str(minivid_crf), '-preset', 'slow', '-pix_fmt', 'yuv444p', osp.join(vid_output, "mini_%06d.mkv" % (vid_counter,))]
print(ffmpeg_proc)
ffmpeg_proc = subprocess.Popen(cmd)#, stderr=subprocess.DEVNULL, stdout=subprocess.DEVNULL)
vid_counter += 1
frame_counter = 0
print("Done.")
if want_just_images:
continue