307 lines
16 KiB
Python
307 lines
16 KiB
Python
#
|
|
# Copyright (C) 2023, Inria
|
|
# GRAPHDECO research group, https://team.inria.fr/graphdeco
|
|
# All rights reserved.
|
|
#
|
|
# This software is free for non-commercial, research and evaluation use
|
|
# under the terms of the LICENSE.md file.
|
|
#
|
|
# For inquiries contact george.drettakis@inria.fr
|
|
#
|
|
|
|
import os
|
|
import torch
|
|
import torch.nn.functional as F
|
|
from random import randint
|
|
from utils.loss_chamfer_utils import chamfer_distance
|
|
from utils.loss_utils import l1_loss, ce_loss, or_loss, ssim
|
|
from gaussian_renderer import render_hair, network_gui
|
|
import sys
|
|
import yaml
|
|
from scene import Scene, GaussianModel, GaussianModelCurves
|
|
from utils.general_utils import safe_state
|
|
import uuid
|
|
from tqdm import tqdm
|
|
from utils.image_utils import psnr, vis_orient
|
|
from argparse import ArgumentParser, Namespace
|
|
from arguments import ModelParams, PipelineParams, OptimizationParams
|
|
import pickle as pkl
|
|
from utils.general_utils import build_rotation
|
|
import time
|
|
from kaolin.metrics.pointcloud import sided_distance
|
|
|
|
try:
|
|
from torch.utils.tensorboard import SummaryWriter
|
|
TENSORBOARD_FOUND = True
|
|
except ImportError:
|
|
TENSORBOARD_FOUND = False
|
|
|
|
|
|
|
|
def training(dataset, opt, opt_hair, pipe, testing_iterations, saving_iterations, checkpoint_iterations, model_path_curves, pointcloud_path_head, checkpoint, checkpoint_hair, debug_from):
|
|
first_iter = 0
|
|
tb_writer = prepare_output_and_logger(dataset, model_path_curves)
|
|
gaussians = GaussianModel(dataset.sh_degree)
|
|
gaussians_hair = GaussianModelCurves(dataset.source_path, dataset.flame_mesh_dir, opt_hair, dataset.sh_degree)
|
|
scene = Scene(dataset, gaussians, pointcloud_path=pointcloud_path_head, load_iteration=-1)
|
|
gaussians.training_setup(opt)
|
|
model_params, _ = torch.load(checkpoint_hair)
|
|
gaussians_hair.create_from_pcd(dataset.source_path, model_params, 30_000, gaussians.spatial_lr_scale)
|
|
gaussians_hair.training_setup(opt)
|
|
if checkpoint:
|
|
model_params, _ = torch.load(checkpoint)
|
|
gaussians.restore(model_params, opt)
|
|
if dataset.trainable_cameras:
|
|
print(f'Loading optimized cameras from iter {scene.loaded_iter}')
|
|
params_cam_rotation, params_cam_translation, params_cam_fov = pkl.load(open(scene.model_path + "/cameras/" + str(scene.loaded_iter) + ".pkl", 'rb'))
|
|
for k in scene.train_cameras.keys():
|
|
for camera in scene.train_cameras[k]:
|
|
if dataset.trainable_cameras:
|
|
camera._rotation_res.data = params_cam_rotation[camera.image_name]
|
|
camera._translation_res.data = params_cam_translation[camera.image_name]
|
|
if dataset.trainable_intrinsics:
|
|
camera._fov_res.data = params_cam_fov[camera.image_name]
|
|
|
|
with torch.no_grad():
|
|
# Head gaussians data
|
|
gaussians.mask_precomp = gaussians.get_label[..., 0] < 0.5
|
|
gaussians.xyz_precomp = gaussians.get_xyz[gaussians.mask_precomp].detach()
|
|
gaussians.opacity_precomp = gaussians.get_opacity[gaussians.mask_precomp].detach()
|
|
gaussians.scaling_precomp = gaussians.get_scaling[gaussians.mask_precomp].detach()
|
|
gaussians.rotation_precomp = gaussians.get_rotation[gaussians.mask_precomp].detach()
|
|
gaussians.cov3D_precomp = gaussians.get_covariance(1.0)[gaussians.mask_precomp].detach()
|
|
gaussians.shs_view = gaussians.get_features[gaussians.mask_precomp].detach().transpose(1, 2).view(-1, 3, (gaussians.max_sh_degree + 1)**2)
|
|
|
|
bg_color = [1, 1, 1, 0, 0, 0, 0, 0, 0, 100] if dataset.white_background else [0, 0, 0, 0, 0, 0, 0, 0, 0, 100]
|
|
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)
|
|
|
|
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):
|
|
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_hair(custom_cam, gaussians, gaussians_hair, 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()
|
|
|
|
gaussians_hair.initialize_gaussians_hair()
|
|
gaussians_hair.update_learning_rate(iteration)
|
|
|
|
# Pick a random Camera
|
|
if not viewpoint_stack:
|
|
viewpoint_stack = scene.getTrainCameras().copy()
|
|
viewpoint_cam = viewpoint_stack.pop(randint(0, len(viewpoint_stack)-1))
|
|
|
|
# Render
|
|
if (iteration - 1) == debug_from:
|
|
pipe.debug = True
|
|
|
|
render_pkg = render_hair(viewpoint_cam, gaussians, gaussians_hair, pipe, background)
|
|
|
|
image = render_pkg["render"]
|
|
mask = render_pkg["mask"]
|
|
orient_angle = render_pkg["orient_angle"]
|
|
orient_conf = render_pkg["orient_conf"]
|
|
|
|
# Loss
|
|
gt_image = viewpoint_cam.original_image.cuda()
|
|
gt_mask = viewpoint_cam.original_mask.cuda()
|
|
gt_orient_angle = viewpoint_cam.original_orient_angle.cuda()
|
|
gt_orient_conf = viewpoint_cam.original_orient_conf.cuda()
|
|
|
|
Ll1 = l1_loss(image, gt_image)
|
|
Lssim = (1.0 - ssim(image, gt_image))
|
|
Lmask = l1_loss(mask, gt_mask)
|
|
|
|
orient_weight = torch.ones_like(gt_mask[:1])
|
|
if opt.use_gt_orient_conf: orient_weight = orient_weight * gt_orient_conf
|
|
if not opt.train_orient_conf: orient_conf = None
|
|
Lorient = or_loss(orient_angle, gt_orient_angle, orient_conf, weight=orient_weight, mask=gt_mask[:1])
|
|
|
|
if torch.isnan(Lorient).any(): Lorient = torch.zeros_like(Ll1)
|
|
|
|
Lsds = gaussians_hair.Lsds if hasattr(gaussians_hair, 'Lsds') and gaussians_hair.Lsds is not None else torch.zeros_like(Ll1)
|
|
|
|
loss = (
|
|
Ll1 * opt.lambda_dl1 +
|
|
Lssim * opt.lambda_dssim +
|
|
Lmask * opt.lambda_dmask +
|
|
Lorient * opt.lambda_dorient +
|
|
Lsds * opt.lambda_dsds
|
|
)
|
|
loss.backward()
|
|
|
|
iter_end.record()
|
|
|
|
# Optimizer step
|
|
if iteration < opt.iterations:
|
|
for param in [gaussians_hair._dirs, gaussians_hair._features_dc, gaussians_hair._features_rest]:
|
|
if param.grad is not None and param.grad.isnan().any():
|
|
gaussians_hair.optimizer.zero_grad(set_to_none = True)
|
|
print('NaN during backprop was found, skipping iteration...')
|
|
|
|
gaussians_hair.optimizer.step()
|
|
gaussians_hair.optimizer.zero_grad()
|
|
|
|
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, Lmask, Lorient, Lsds, loss, l1_loss, or_loss, iter_start.elapsed_time(iter_end), testing_iterations, scene, gaussians_hair, render_hair, (pipe, background))
|
|
if (iteration in saving_iterations):
|
|
print("\n[ITER {}] Saving Gaussians".format(iteration))
|
|
scene.save(iteration)
|
|
|
|
if (iteration in checkpoint_iterations):
|
|
print("\n[ITER {}] Saving Checkpoint".format(iteration))
|
|
os.makedirs(model_path_curves + "/checkpoints", exist_ok=True)
|
|
torch.save((gaussians_hair.capture(), iteration), model_path_curves + "/checkpoints/" + str(iteration) + ".pth")
|
|
|
|
def prepare_output_and_logger(args, model_path_curves):
|
|
if not model_path_curves:
|
|
if os.getenv('OAR_JOB_ID'):
|
|
unique_str=os.getenv('OAR_JOB_ID')
|
|
else:
|
|
unique_str = str(uuid.uuid4())
|
|
model_path_curves = os.path.join("./output/", unique_str[0:10])
|
|
|
|
# Set up output folder
|
|
print("Output folder: {}".format(model_path_curves))
|
|
os.makedirs(model_path_curves, exist_ok = True)
|
|
with open(os.path.join(model_path_curves, "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(model_path_curves)
|
|
else:
|
|
print("Tensorboard not available: not logging progress")
|
|
return tb_writer
|
|
|
|
def training_report(tb_writer, iteration, Ll1, Lmask, Lorient, Lsds, loss, l1_loss, or_loss, elapsed, testing_iterations, scene : Scene, gaussians_hair, renderFunc, renderArgs):
|
|
if tb_writer:
|
|
tb_writer.add_scalar('train_loss_patches/l1_loss', Ll1.item(), iteration)
|
|
tb_writer.add_scalar('train_loss_patches/ce_loss', Lmask.item(), iteration)
|
|
tb_writer.add_scalar('train_loss_patches/or_loss', Lorient.item(), iteration)
|
|
tb_writer.add_scalar('train_loss_patches/df_loss', Lsds.item(), iteration)
|
|
tb_writer.add_scalar('train_loss_patches/total_loss', loss.item(), iteration)
|
|
tb_writer.add_scalar('iter_time', elapsed, iteration)
|
|
|
|
# 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:
|
|
Ll1_test = 0.0
|
|
Lmask_test = 0.0
|
|
Lorient_test = 0.0
|
|
psnr_test = 0.0
|
|
for idx, viewpoint in enumerate(config['cameras']):
|
|
render_pkg = renderFunc(viewpoint, scene.gaussians, gaussians_hair, *renderArgs)
|
|
image = torch.clamp(render_pkg["render"], 0.0, 1.0)
|
|
mask = torch.clamp(render_pkg["mask"], 0.0, 1.0)
|
|
orient_angle = torch.clamp(render_pkg["orient_angle"], 0.0, 1.0)
|
|
orient_conf = render_pkg["orient_conf"]
|
|
orient_conf_vis = (1 - 1 / (orient_conf + 1)) * mask[:1]
|
|
gt_image = torch.clamp(viewpoint.original_image.to("cuda"), 0.0, 1.0)
|
|
gt_mask = torch.clamp(viewpoint.original_mask.to("cuda"), 0.0, 1.0)
|
|
gt_orient_angle = torch.clamp(viewpoint.original_orient_angle.to("cuda"), 0.0, 1.0)
|
|
gt_orient_conf = viewpoint.original_orient_conf.to("cuda")
|
|
gt_orient_conf_vis = (1 - 1 / (gt_orient_conf + 1)) * gt_mask[:1]
|
|
if tb_writer and (idx < 5):
|
|
tb_writer.add_images(config['name'] + "_view_{}/render".format(viewpoint.image_name), image[None], global_step=iteration)
|
|
tb_writer.add_images(config['name'] + "_view_{}/render_mask".format(viewpoint.image_name), F.pad(mask, (0, 0, 0, 0, 0, 3-mask.shape[0]), 'constant', 0)[None], global_step=iteration)
|
|
tb_writer.add_images(config['name'] + "_view_{}/render_orient".format(viewpoint.image_name), vis_orient(orient_angle, mask[:1])[None], global_step=iteration)
|
|
tb_writer.add_images(config['name'] + "_view_{}/render_orient_conf".format(viewpoint.image_name), vis_orient(orient_angle, orient_conf_vis)[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)
|
|
tb_writer.add_images(config['name'] + "_view_{}/ground_truth_mask".format(viewpoint.image_name), F.pad(gt_mask, (0, 0, 0, 0, 0, 3-gt_mask.shape[0]), 'constant', 0)[None], global_step=iteration)
|
|
tb_writer.add_images(config['name'] + "_view_{}/ground_truth_orient".format(viewpoint.image_name), vis_orient(gt_orient_angle, gt_mask[:1])[None], global_step=iteration)
|
|
tb_writer.add_images(config['name'] + "_view_{}/ground_truth_orient_conf".format(viewpoint.image_name), vis_orient(gt_orient_angle, gt_orient_conf_vis)[None], global_step=iteration)
|
|
Ll1_test += l1_loss(image, gt_image).mean().double()
|
|
Lmask_test += l1_loss(mask, gt_mask).mean().double()
|
|
Lorient_test += or_loss(orient_angle, gt_orient_angle, mask=gt_mask[:1], weight=gt_orient_conf).mean().double()
|
|
psnr_test += psnr(image, gt_image).mean().double()
|
|
Ll1_test /= len(config['cameras'])
|
|
Lmask_test /= len(config['cameras'])
|
|
Lorient_test /= len(config['cameras'])
|
|
psnr_test /= len(config['cameras'])
|
|
print("\n[ITER {}] Evaluating {}: L1 {} CE {} OR {} PSNR {}".format(iteration, config['name'], Ll1_test, Lmask_test, Lorient_test, psnr_test))
|
|
if tb_writer:
|
|
tb_writer.add_scalar(config['name'] + '/loss_viewpoint - l1_loss', Ll1_test, iteration)
|
|
tb_writer.add_scalar(config['name'] + '/loss_viewpoint - ce_loss', Lmask_test, iteration)
|
|
tb_writer.add_scalar(config['name'] + '/loss_viewpoint - or_loss', Lorient_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.gaussians.get_opacity, iteration)
|
|
tb_writer.add_histogram("scene/label_histogram", scene.gaussians.get_label, iteration)
|
|
tb_writer.add_scalar('total_points', scene.gaussians.get_xyz.shape[0], iteration)
|
|
torch.cuda.empty_cache()
|
|
|
|
if __name__ == "__main__":
|
|
# Set up command line argument parser
|
|
parser = ArgumentParser(description="Training script parameters")
|
|
lp = ModelParams(parser)
|
|
op = OptimizationParams(parser)
|
|
pp = PipelineParams(parser)
|
|
parser.add_argument('--ip', type=str, default="127.0.0.1")
|
|
parser.add_argument('--port', type=int, default=6009)
|
|
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=[1_000, 5_000, 10_000])
|
|
parser.add_argument("--save_iterations", nargs="+", type=int, default=[1_000, 5_000, 10_000])
|
|
parser.add_argument("--quiet", action="store_true")
|
|
parser.add_argument("--checkpoint_iterations", nargs="+", type=int, default=[1_000, 5_000, 10_000])
|
|
parser.add_argument("--start_checkpoint", type=str, default = None)
|
|
parser.add_argument("--start_checkpoint_hair", type=str, default = None)
|
|
parser.add_argument("--hair_conf_path", type=str, default = None)
|
|
parser.add_argument("--model_path_curves", type=str, default = None)
|
|
parser.add_argument("--pointcloud_path_head", type=str, default = None)
|
|
args = parser.parse_args(sys.argv[1:])
|
|
args.save_iterations.append(args.iterations)
|
|
|
|
print("Optimizing " + args.model_path_curves)
|
|
|
|
# Configuration of hair strands
|
|
with open(args.hair_conf_path, 'r') as f:
|
|
replaced_conf = str(yaml.load(f, Loader=yaml.Loader)).replace('DATASET_TYPE', 'monocular')
|
|
opt_hair = yaml.load(replaced_conf, Loader=yaml.Loader)
|
|
|
|
# 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(lp.extract(args), op.extract(args), opt_hair, pp.extract(args), args.test_iterations, args.save_iterations, args.checkpoint_iterations, args.model_path_curves, args.pointcloud_path_head, args.start_checkpoint, args.start_checkpoint_hair, args.debug_from)
|
|
|
|
# All done
|
|
print("\nTraining complete.")
|