-
Notifications
You must be signed in to change notification settings - Fork 2
/
train.py
279 lines (257 loc) · 14.3 KB
/
train.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
#
# Copyright (C) 2023, Inria
# GRAPHDECO research group, https://team.inria.fr/graphdeco
# All rights reserved.
# For inquiries contact george.drettakis@inria.fr
# Copyright (C) 2024, Jiajie Yang https://github.com/RANDO11199
# All rights reserved.
# This software is free for non-commercial, research and evaluation use
# under the terms of the LICENSE.md file.
#
#
import warnings
warnings.filterwarnings('ignore')
from torchvision.utils import save_image,make_grid
import os
import torch
from pytorch3d.io import save_ply
from random import randint
from renderer import Rasterize
from utils.loss_utils import l1_loss, ssim,tv_loss
from pytorch3d.loss import mesh_edge_loss,mesh_laplacian_smoothing,mesh_normal_consistency
from renderer import SHrender,network_gui,RayTracer
from renderer.render_utils import Init_shaer
import sys
from scene import Scene,TripletModel
from utils.general_utils import safe_state
import uuid
from tqdm import tqdm
from torch.nn.functional import interpolate
from utils.loss_utils import psnr
from argparse import ArgumentParser, Namespace
from arguements import ModelParams, PipelineParams, OptimizationParams, LightParams, align_paramater_relfectionmodel, align_paramater_scene_type
from functools import partial
torch.backends.cudnn.enabled = True
torch.backends.cudnn.benchmark = True
try:
from torch.utils.tensorboard import SummaryWriter
TENSORBOARD_FOUND = True
except ImportError:
TENSORBOARD_FOUND = False
def training(dataset, lightset, opt, pipe, testing_iterations, saving_iterations, checkpoint_iterations, checkpoint, debug_from):
first_iter = 0
tb_writer = prepare_output_and_logger(dataset)
triplet = TripletModel()
scene = Scene(dataset, lightset, triplet)
triplet.training_setup(opt)
shader = Init_shaer(dataset)
if checkpoint:
(model_params,scene_params, first_iter) = torch.load(checkpoint)
triplet.restore(model_params, opt)
scene.restore(scene_params,lightset)
bg_color = [1, 1, 1] if dataset.white_background else [0, 0, 0]
background = torch.tensor(bg_color, dtype=torch.float32, device="cuda")
iter_start = torch.cuda.Event(enable_timing = True)
iter_end = torch.cuda.Event(enable_timing = True)
if dataset.renderer == 'SH':
render = SHrender
elif dataset.renderer == 'Rasterization':
render = partial(Rasterize,shader)
elif dataset.renderer == 'Whitted-StyleRT':
render = RayTracer
else:
raise NotImplementedError(f'Renderer {dataset.renderer} is not implemented yet')
viewpoint_stack = None
ema_loss_for_log = 0.0
progress_bar = tqdm(range(first_iter, opt.iterations), desc="Training progress")
first_iter += 1
for iteration in range(first_iter, opt.iterations + 1):
triplet.check_progress(iteration,opt)
scene.check_progress(iteration)
# TODO: Currently not supported
if network_gui.conn == None:
network_gui.try_connect()
while network_gui.conn != None:
try:
net_image_bytes = None
custom_cam, do_training, pipe.convert_SHs_python, pipe.compute_cov3D_python, keep_alive, scaling_modifer = network_gui.receive()
if custom_cam != None:
net_image = render(shader,custom_cam, gaussians, pipe, background, scaling_modifer)["render"]
net_image_bytes = memoryview((torch.clamp(net_image, min=0, max=1.0) * 255).byte().permute(1, 2, 0).contiguous().cpu().numpy())
network_gui.send(net_image_bytes, dataset.source_path)
if do_training and ((iteration < int(opt.iterations)) or not keep_alive):
break
except Exception as e:
network_gui.conn = None
iter_start.record()
triplet.update_learning_rate(iteration)
# Pick a random Camera
if not viewpoint_stack:
viewpoint_stack = scene.getTrainCameras().copy()
rand_num = randint(0, len(viewpoint_stack)-1)
viewpoint_cam = viewpoint_stack.pop(rand_num)
light,diffuse_map,specular_map = scene.getLightSource()
# Render
if (iteration - 1) == debug_from:
pipe.debug = True
bg = torch.rand((3), device="cuda") if opt.random_background else background
# with torch.autograd.profiler.profile(enabled=True,use_cuda=True,with_flops=True,profile_memory=True,use_cpu=False,use_kineto=True) as prof:
render_pkg = render(target_camera = viewpoint_cam, light = light, triplet = triplet, pipe = pipe, bg_color = bg,scaling=triplet.resolution_scale)
# print(prof.key_averages().table(sort_by="self_cuda_time_total"))
image, visibility_filter, area, depth_map, normal_map,ndc_grad = render_pkg["render"], render_pkg["visibility_filter"], render_pkg["area"],render_pkg['depth_map'],render_pkg['normal_map'],render_pkg['ndc_grad']
# Loss
gt_image = viewpoint_cam.original_image[None]
gt_image = interpolate(gt_image,scale_factor=triplet.resolution_scale)[0]
gt_mask = viewpoint_cam.gt_alpha_mask
if gt_mask is not None :
gt_mask = interpolate(gt_mask[None],scale_factor=triplet.resolution_scale)[0]
gt_image = gt_image + torch.ones_like(gt_image,device=gt_image.device) * bg.reshape(3,1,1) * ( 1- gt_mask)
Ll1 = l1_loss(image, gt_image,reduce='mean')
loss = (1.0 - opt.lambda_dssim) * (Ll1 ) + opt.lambda_dssim *(1.0 - ssim(image, gt_image))
if iteration >= 7000:
loss += (0.01 * tv_loss(image[None]) +
0.05 * tv_loss(render_pkg['pseudo_normal'][None])
) # Local Invariants/Continuous
loss += 0.05 * l1_loss(render_pkg['pseudo_normal'][None],normal_map,reduce='mean')
loss.backward()
torch.nn.utils.clip_grad.clip_grad_norm_(triplet._feature_rest_s,1.)
iter_end.record()
with torch.no_grad():
# Progress bar
ema_loss_for_log = 0.4 * loss.item() + 0.6 * ema_loss_for_log
if iteration % 10 == 0:
progress_bar.set_postfix({"Loss": f"{ema_loss_for_log:.{7}f}"})
progress_bar.update(10)
if iteration == opt.iterations:
progress_bar.close()
# Log and save
training_report(tb_writer, iteration, Ll1, loss, l1_loss, iter_start.elapsed_time(iter_end), testing_iterations, scene, render, (pipe, background),
render_img=image,gt_image = gt_image,normal_image =render_pkg['normal_map'],depth_image=depth_map,opt=dataset,pseudo_normal = render_pkg['pseudo_normal'])
if (iteration in saving_iterations):
print("\n[ITER {}] Saving Triplet".format(iteration))
scene.save(iteration)
triplet.max_screen_area[visibility_filter[0]] = torch.max(triplet.max_screen_area[visibility_filter[0]],area)
# Kill large mesh on screen
if iteration < opt.densify_from_iter or iteration % opt.densification_interval == 0:
triplet.kill_large_mesh(max_screen_size = gt_image.shape[1]*gt_image.shape[2] * opt.screen_max )
# Densification
if iteration < opt.densify_until_iter:
# Keep track of max radii in image-space for pruning
triplet.add_densification_stats(visibility_filter[1],ndc_grad)
if iteration > opt.densify_from_iter and iteration % opt.densification_interval == 0:
triplet.densify_and_prune(opt.densify_grad_threshold, 0.05, scene.cameras_extent)
if iteration % opt.opacity_reset_interval == 0 or (dataset.white_background and iteration == opt.densify_from_iter):
triplet.reset_alpha()
# Optimizer step
if iteration < opt.iterations:
triplet.optimizer.step()
triplet.optimizer.zero_grad(set_to_none=True)
scene.optimizer.step()
scene.optimizer.zero_grad(set_to_none=True)
if (iteration in checkpoint_iterations):
print("\n[ITER {}] Saving Checkpoint".format(iteration))
torch.save((triplet.capture(),scene.capture() ,iteration), scene.model_path + "/chkpnt" + str(iteration) + ".pth")
# triplet.setup_view_dependent_compensation(iteration,opt.view_compensation_from_iter)
# scene.check_grad_progress(iteration,opt)
def prepare_output_and_logger(args):
if not args.model_path:
if os.getenv('OAR_JOB_ID'):
unique_str=os.getenv('OAR_JOB_ID')
else:
unique_str = str(uuid.uuid4())
args.model_path = os.path.join("./output/", unique_str[0:10])
# Set up output folder
print("Output folder: {}".format(args.model_path))
os.makedirs(args.model_path, exist_ok = True)
with open(os.path.join(args.model_path, "cfg_args"), 'w') as cfg_log_f:
cfg_log_f.write(str(Namespace(**vars(args))))
# Create Tensorboard writer
tb_writer = None
if TENSORBOARD_FOUND:
tb_writer = SummaryWriter(args.model_path)
else:
print("Tensorboard not available: not logging progress")
return tb_writer
def training_report(tb_writer:SummaryWriter, iteration, Ll1, loss, l1_loss, elapsed, testing_iterations, scene : Scene, renderFunc, renderArgs,**kwargs):
if tb_writer:
tb_writer.add_scalar('train_loss_patches/l1_loss', Ll1.item(), iteration)
tb_writer.add_scalar('train_loss_patches/total_loss', loss.item(), iteration)
tb_writer.add_scalar('iter_time', elapsed, iteration)
if iteration % 100 == 0 :
# scene.triplets.save_ply_geometry(iteration)
try:
sav_img = torch.stack([kwargs['render_img'],kwargs['gt_image'],
kwargs['normal_image'][0],(kwargs['pseudo_normal']),
torch.from_numpy(kwargs['depth_image']).permute(2,0,1).to('cuda')
],dim =0)
sav_img = make_grid(sav_img,nrow=2)
save_image(sav_img,f'./output/test_{iteration}.png')
except:
print('saving image fail')
# Report test and samples of training set
if iteration in testing_iterations:
torch.cuda.empty_cache()
validation_configs = ({'name': 'test', 'cameras' : scene.getTestCameras()},
{'name': 'train', 'cameras' : [scene.getTrainCameras()[idx % len(scene.getTrainCameras())] for idx in range(5, 30, 5)]})
for config in validation_configs:
if config['cameras'] and len(config['cameras']) > 0:
l1_test = 0.0
psnr_test = 0.0
for idx, viewpoint in enumerate(config['cameras']):
lightsource,_,_ = scene.getLightSource()
result = renderFunc(target_camera = viewpoint, light = lightsource,bg_color = renderArgs[1],pipe=None,triplet= scene.triplets)["render"]
image = torch.clamp(result, 0.0, 1.0)
gt_image = torch.clamp(viewpoint.original_image.to("cuda"), 0.0, 1.0)
if viewpoint.gt_alpha_mask is not None:
gt_image = gt_image + torch.ones_like(gt_image).cuda() * ( 1- viewpoint.gt_alpha_mask)
if tb_writer and (idx < 5):
tb_writer.add_images(config['name'] + "_view_{}/render".format(viewpoint.image_name), image[None], global_step=iteration)
if iteration == testing_iterations[0]:
tb_writer.add_images(config['name'] + "_view_{}/ground_truth".format(viewpoint.image_name), gt_image[None], global_step=iteration)
l1_test += l1_loss(image, gt_image).mean().double()
psnr_test += psnr(image, gt_image).mean().double()
psnr_test /= len(config['cameras'])
l1_test /= len(config['cameras'])
print("\n[ITER {}] Evaluating {}: L1 {} PSNR {}".format(iteration, config['name'], l1_test, psnr_test))
if tb_writer:
tb_writer.add_scalar(config['name'] + '/loss_viewpoint - l1_loss', l1_test, iteration)
tb_writer.add_scalar(config['name'] + '/loss_viewpoint - psnr', psnr_test, iteration)
if tb_writer:
tb_writer.add_histogram("scene/opacity_histogram", scene.triplets.get_alpha, iteration)
tb_writer.add_scalar('total_points', scene.triplets.get_deform.shape[1], iteration)
torch.cuda.empty_cache()
if __name__ == "__main__":
# Set up command line argument parser
parser = ArgumentParser(description="Training script parameters")
mp = ModelParams(parser)
op = OptimizationParams(parser)
pp = PipelineParams(parser)
lp = LightParams(parser)
parser.add_argument('--ip', type=str, default="127.0.0.1")
parser.add_argument('--port', type=int, default=6019)
parser.add_argument('--debug_from', type=int, default=-1)
parser.add_argument('--detect_anomaly', action='store_true', default=False)
parser.add_argument("--test_iterations", nargs="+", type=int, default=[7000, 14000,16000, 18000,19000,20000, 21000,22000,23000,24000,25000,30_000])
parser.add_argument("--save_iterations", nargs="+", type=int, default=[])# Come back later
parser.add_argument("--quiet", action="store_true")
parser.add_argument("--checkpoint_iterations", nargs="+", type=int, default=[30_000])
parser.add_argument("--start_checkpoint", type=str, default = None)
parser.add_argument('--align_param', action='store_true', default=True)
args = parser.parse_args(sys.argv[1:])
args.save_iterations.append(args.iterations)
print(lp.compensate_random_Point)
lp = lp.extract(args)
mp = mp.extract(args)
op = op.extract(args)
if args.align_param:
lp,op = align_paramater_relfectionmodel(lp,op,mp)
lp,op = align_paramater_scene_type(lp,op,mp)
print("Optimizing " + args.model_path)
# Initialize system state (RNG)
safe_state(args.quiet)
# Start GUI server, configure and run training
network_gui.init(args.ip, args.port)
torch.autograd.set_detect_anomaly(args.detect_anomaly)
training(mp,lp, op, pp.extract(args), args.test_iterations, args.save_iterations, args.checkpoint_iterations, args.start_checkpoint, args.debug_from)
# All done
print("\nTraining complete.")