mirror of
https://github.com/nerfstudio-project/nerfstudio.git
synced 2026-11-11 18:44:21 +08:00
0b042be956
* Model class done, pipeline class still needs more visualization stuffs * Renaming the BaseDataContainer to DataloaderOutputs and adding documentation * Pushing more documentation, removing some trivial functions * Updating dict typing * Removing config dicts for collider * Passing tests * Pushing shell new dataloader * Saving work, need to fix NGP model's get_loss_dict and Pipeline * V1 of the ngp pipeline should be done. Need train script and debugging * Updating with latest instant-ngp code * Syntax updates from latest ngp graph code * Linter, removing unneeded code in new NGP graph code * Style * Should be good now. Linter is happy and the code is verbatum current instant-ngp * Everything done except novel view testing code * Fixing some nitpicks * Debugging dataloader * merged master * WOP. integrating pipeline with configs and notebook demo * removing model folder * a lot of name changing from Graph to Model * in progress... changing trainer to use pipeline... broken right now * pipeline code runs * switched to val dataset * makes run_eval.py work * updated configs for pipeline change * fixing test_run_train script * fixing black and linting * fix docs * small updates to trainer Co-authored-by: Jake Austin <jake-austin@berkeley.edu>
124 lines
5.2 KiB
Python
124 lines
5.2 KiB
Python
# Copyright 2022 The Plenoptix Team. All rights reserved.
|
|
#
|
|
# Licensed under the Apache License, Version 2.0 (the "License");
|
|
# you may not use this file except in compliance with the License.
|
|
# You may obtain a copy of the License at
|
|
#
|
|
# http://www.apache.org/licenses/LICENSE-2.0
|
|
#
|
|
# Unless required by applicable law or agreed to in writing, software
|
|
# distributed under the License is distributed on an "AS IS" BASIS,
|
|
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
# See the License for the specific language governing permissions and
|
|
# limitations under the License.
|
|
|
|
"""
|
|
Semantic NeRF implementation.
|
|
"""
|
|
|
|
import torch
|
|
from torch import nn
|
|
|
|
from nerfactory.cameras.rays import RayBundle
|
|
from nerfactory.dataloaders.structs import Semantics
|
|
from nerfactory.fields.modules.encoding import NeRFEncoding
|
|
from nerfactory.fields.modules.field_heads import FieldHeadNames
|
|
from nerfactory.fields.nerf_field import NeRFField
|
|
from nerfactory.fields.semantic_nerf_field import SemanticNerfField
|
|
from nerfactory.models.vanilla_nerf import NeRFModel
|
|
from nerfactory.renderers.renderers import SemanticRenderer
|
|
from nerfactory.utils import misc, writer
|
|
|
|
|
|
class SemanticNerfModel(NeRFModel):
|
|
"""Semantic-NeRF model"""
|
|
|
|
def __init__(self, semantics: Semantics, **kwargs) -> None:
|
|
self.stuff_classes = semantics.stuff_classes
|
|
self.stuff_colors = semantics.stuff_colors
|
|
super().__init__(**kwargs)
|
|
|
|
def populate_fields(self):
|
|
"""Set the fields."""
|
|
position_encoding = NeRFEncoding(
|
|
in_dim=3, num_frequencies=10, min_freq_exp=0.0, max_freq_exp=8.0, include_input=True
|
|
)
|
|
direction_encoding = NeRFEncoding(
|
|
in_dim=3, num_frequencies=4, min_freq_exp=0.0, max_freq_exp=4.0, include_input=True
|
|
)
|
|
self.field_coarse = NeRFField(position_encoding=position_encoding, direction_encoding=direction_encoding)
|
|
|
|
num_semantic_classes = len(self.stuff_classes)
|
|
self.field_fine = SemanticNerfField(
|
|
num_semantic_classes, position_encoding=position_encoding, direction_encoding=direction_encoding
|
|
)
|
|
|
|
def populate_misc_modules(self):
|
|
super().populate_misc_modules()
|
|
self.renderer_semantic = SemanticRenderer()
|
|
self.cross_entropy_loss = nn.CrossEntropyLoss(reduction="mean")
|
|
|
|
def get_outputs(self, ray_bundle: RayBundle):
|
|
# uniform sampling
|
|
ray_samples_uniform = self.sampler_uniform(ray_bundle)
|
|
|
|
# coarse field
|
|
field_outputs_coarse = self.field_coarse.forward(ray_samples_uniform)
|
|
weights_coarse = ray_samples_uniform.get_weights(field_outputs_coarse[FieldHeadNames.DENSITY])
|
|
rgb_coarse = self.renderer_rgb(
|
|
rgb=field_outputs_coarse[FieldHeadNames.RGB],
|
|
weights=weights_coarse,
|
|
)
|
|
accumulation_coarse = self.renderer_accumulation(weights_coarse)
|
|
depth_coarse = self.renderer_depth(weights_coarse, ray_samples_uniform)
|
|
|
|
# pdf sampling
|
|
ray_samples_pdf = self.sampler_pdf(ray_bundle, ray_samples_uniform, weights_coarse)
|
|
|
|
# fine field
|
|
field_outputs_fine = self.field_fine.forward(ray_samples_pdf)
|
|
weights_fine = ray_samples_pdf.get_weights(field_outputs_fine[FieldHeadNames.DENSITY])
|
|
rgb_fine = self.renderer_rgb(
|
|
rgb=field_outputs_fine[FieldHeadNames.RGB],
|
|
weights=weights_fine,
|
|
)
|
|
accumulation_fine = self.renderer_accumulation(weights_fine)
|
|
depth_fine = self.renderer_depth(weights_fine, ray_samples_pdf)
|
|
semantic_fine = self.renderer_semantic(field_outputs_fine[FieldHeadNames.SEMANTICS], weights=weights_fine)
|
|
|
|
outputs = {
|
|
"rgb_coarse": rgb_coarse,
|
|
"rgb_fine": rgb_fine,
|
|
"accumulation_coarse": accumulation_coarse,
|
|
"accumulation_fine": accumulation_fine,
|
|
"depth_coarse": depth_coarse,
|
|
"depth_fine": depth_fine,
|
|
"semantic_fine": semantic_fine,
|
|
}
|
|
return outputs
|
|
|
|
def get_loss_dict(self, outputs, batch, metrics_dict, loss_coefficients):
|
|
image = batch["image"]
|
|
rgb_loss_coarse = self.rgb_loss(image, outputs["rgb_coarse"])
|
|
rgb_loss_fine = self.rgb_loss(image, outputs["rgb_fine"])
|
|
semantic_logits = outputs["semantic_fine"]
|
|
semantic_classes = batch["semantics"][..., 0].long()
|
|
semantic_loss_fine = self.cross_entropy_loss(semantic_logits, semantic_classes)
|
|
loss_dict = {
|
|
"rgb_loss_coarse": rgb_loss_coarse,
|
|
"rgb_loss_fine": rgb_loss_fine,
|
|
"semantic_loss_fine": semantic_loss_fine,
|
|
}
|
|
loss_dict = misc.scale_dict(loss_dict, loss_coefficients)
|
|
return loss_dict
|
|
|
|
def log_test_image_outputs(self, image_idx, step, batch, outputs):
|
|
super().log_test_image_outputs(image_idx, step, batch, outputs)
|
|
semantic_logits = outputs["semantic_fine"]
|
|
semantic_labels = torch.argmax(torch.nn.functional.softmax(semantic_logits, dim=-1), dim=-1) # type: ignore
|
|
semantic_labels_image = self.stuff_colors[semantic_labels]
|
|
writer.put_image(name=f"semantics/image_idx_{image_idx}", image=semantic_labels_image, step=step)
|
|
|
|
mask = batch["mask"].repeat(1, 1, 3)
|
|
writer.put_image(name=f"mask/image_idx_{image_idx}", image=mask, step=step)
|