Add more samplers

This commit is contained in:
Matt Tancik authored and Matthew Tancik committed 2022-06-16 18:27:47 -07:00
1 parent 025648e729
commit 65fe212d17
5 files changed
+375 -38

No files matched your search

+9 -1
View File
@@ -14,4 +14,12 @@ Encoders
.. toctree::
:maxdepth: 0
visualize_encoders.ipynb
visualize_encoders.ipynb
Point Samplers
----------------
.. toctree::
:maxdepth: 0
visualize_samplers.ipynb
+154
View File
@@ -0,0 +1,154 @@
{
"cells": [
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Visualize Samplers"
]
},
{
"cell_type": "code",
"execution_count": 1,
"metadata": {},
"outputs": [],
"source": [
"import torch\n",
"from pyrad.cameras.rays import RayBundle\n",
"from pyrad.graphs.modules import ray_sampler\n",
"from matplotlib import pyplot as plt"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Spaced Samplers\n",
"\n",
"A basic sampler that spaces samples based on a predefined function."
]
},
{
"cell_type": "code",
"execution_count": 2,
"metadata": {},
"outputs": [
{
"data": {
"image/png": "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 truncated
"text/plain": [
"<Figure size 432x288 with 1 Axes>"
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
},
{
"data": {
"image/png": "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 truncated
"text/plain": [
"<Figure size 432x288 with 1 Axes>"
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
},
{
"data": {
"image/png": "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 truncated
"text/plain": [
"<Figure size 432x288 with 1 Axes>"
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
},
{
"data": {
"image/png": "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 truncated
"text/plain": [
"<Figure size 432x288 with 1 Axes>"
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
}
],
"source": [
"num_samples = 1000\n",
"near = 2\n",
"far = 5\n",
"train_stratified = False\n",
"\n",
"samplers = {\n",
" \"Uniform\": ray_sampler.UniformSampler,\n",
" \"Linear in Disparity\": ray_sampler.LinearDisparitySampler,\n",
" \"Square Root\": ray_sampler.SqrtSampler,\n",
" \"Log Sampler\": ray_sampler.LogSampler,\n",
"}\n",
"\n",
"for name, Sampler in samplers.items():\n",
" sampler = Sampler(num_samples=num_samples, train_stratified=train_stratified)\n",
"\n",
" ray_bundle = RayBundle(\n",
" origins=torch.ones([1, 3]),\n",
" directions=torch.ones([1, 3]),\n",
" pixel_area=torch.ones([1, 1]),\n",
" nears=torch.ones([1]) * near,\n",
" fars=torch.ones([1]) * far,\n",
" )\n",
"\n",
" samples = sampler.generate_ray_samples(ray_bundle)\n",
" samples.bin_starts.shape\n",
"\n",
" plt.hist(samples.bin_starts, alpha=0.5, bins=100)\n",
" plt.xlabel(\"Distance\")\n",
" plt.ylabel(\"# Samples\")\n",
" plt.title(name)\n",
" plt.show()"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": []
}
],
"metadata": {
"kernelspec": {
"display_name": "Python 3.8.12 ('pyrad')",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.8.12"
},
"orig_nbformat": 4,
"vscode": {
"interpreter": {
"hash": "41f62990c6e03ccb48eec29860b83c64e20fcd45204ae5014afab6809cd6f1f1"
}
}
},
"nbformat": 4,
"nbformat_minor": 2
}
+2 -2
View File
@@ -210,8 +210,8 @@ class RayBundle:
directions (TensorType[..., 3]): Unit ray direction vector
pixel_area (TensorType[..., 1]): Projected area of pixel a distance 1 away from origin.
camera_indices (TensorType[..., 1]): Camera indices
nears (TensorType[..., 1]): Distance along ray to start sampling
fars (TensorType[..., 1]): Rays Distance along ray to stop sampling
nears (TensorType[...]): Distance along ray to start sampling
fars (TensorType[...]): Rays Distance along ray to stop sampling
valid_mask (TensorType[...]): Rays that are valid
"""
+152 -34
View File
@@ -17,7 +17,7 @@ Collection of sampling strategies
"""
from abc import abstractmethod
from typing import Optional, Tuple
from typing import Callable, Optional, Tuple
import torch
from torch import nn
@@ -54,7 +54,73 @@ class Sampler(nn.Module):
return ray_samples
class UniformSampler(Sampler):
class SpacedSampler(Sampler):
"""Sample points according to a function."""
def __init__(
self,
num_samples: int,
spacing_fn: Callable,
spacing_fn_inv: Callable,
train_stratified=True,
occupancy_field: Optional[OccupancyGrid] = None,
weight_threshold: float = 1e-4,
) -> None:
"""
Args:
num_samples (int): Number of samples per ray
spacing_fn (Callable): Function that dictates sample spacing (ie `lambda x : x` is uniform).
spacing_fn_inv (Callable): The inverse of spacing_fn.
train_stratified (bool): Use stratified sampling during training. Defults to True
occupancy_field (OccupancyGrid, optional): Occupancy grid. If provides,
samples below weight_threshold as set as invalid.
weight_thershold (float): Removes samples below threshold weight. Only used if occupancy field is provided.
"""
super().__init__(num_samples=num_samples, occupancy_field=occupancy_field, weight_threshold=weight_threshold)
self.train_stratified = train_stratified
self.spacing_fn = spacing_fn
self.spacing_fn_inv = spacing_fn_inv
@torch.no_grad()
def generate_ray_samples(
self,
ray_bundle: RayBundle = None,
num_samples: Optional[int] = None,
) -> RaySamples:
"""Generates position samples accoring to spacing function.
Args:
ray_bundle (RayBundle): Rays to generate samples for
num_samples (Optional[int]): Number of samples per ray
Returns:
RaySamples: Positions and deltas for samples along a ray
"""
assert ray_bundle is not None
assert ray_bundle.nears is not None
assert ray_bundle.fars is not None
num_samples = num_samples or self.num_samples
num_rays = ray_bundle.origins.shape[0]
bins = torch.linspace(0.0, 1.0, num_samples + 1).to(ray_bundle.origins.device)[None, ...] # [1, num_samples+1]
s_near, s_far = [self.spacing_fn(x) for x in (ray_bundle.nears[:, None], ray_bundle.fars[:, None])]
bins = self.spacing_fn_inv(bins * s_far + (1 - bins) * s_near) # [num_rays, num_samples+1]
if self.train_stratified and self.training:
t_rand = torch.rand((num_rays, num_samples + 1), dtype=bins.dtype, device=bins.device)
bin_centers = (bins[..., 1:] + bins[..., :-1]) / 2.0
bin_upper = torch.cat([bin_centers, bins[..., -1:]], -1)
bin_lower = torch.cat([bins[..., :1], bin_centers], -1)
bins = bin_lower + (bin_upper - bin_lower) * t_rand
ray_samples = ray_bundle.get_ray_samples(bin_starts=bins[..., :-1], bin_ends=bins[..., 1:])
return ray_samples
class UniformSampler(SpacedSampler):
"""Sample uniformly along a ray"""
def __init__(
@@ -72,46 +138,98 @@ class UniformSampler(Sampler):
samples below weight_threshold as set as invalid.
weight_thershold (float): Removes samples below threshold weight. Only used if occupancy field is provided.
"""
super().__init__(num_samples=num_samples, occupancy_field=occupancy_field, weight_threshold=weight_threshold)
self.train_stratified = train_stratified
super().__init__(
num_samples=num_samples,
spacing_fn=lambda x: x,
spacing_fn_inv=lambda x: x,
train_stratified=train_stratified,
occupancy_field=occupancy_field,
weight_threshold=weight_threshold,
)
@torch.no_grad()
def generate_ray_samples(
class LinearDisparitySampler(SpacedSampler):
"""Sample linearly in disparity along a ray"""
def __init__(
self,
ray_bundle: RayBundle = None,
num_samples: Optional[int] = None,
) -> RaySamples:
"""Generates position samples uniformly.
Args:
ray_bundle (RayBundle): Rays to generate samples for
num_samples (Optional[int]): Number of samples per ray
Returns:
RaySamples: Positions and deltas for samples along a ray
num_samples: int,
train_stratified=True,
occupancy_field: Optional[OccupancyGrid] = None,
weight_threshold: float = 1e-4,
) -> None:
"""
assert ray_bundle is not None
assert ray_bundle.nears is not None
assert ray_bundle.fars is not None
Args:
num_samples (int): Number of samples per ray
train_stratified (bool): Use stratified sampling during training. Defults to True
occupancy_field (OccupancyGrid, optional): Occupancy grid. If provides,
samples below weight_threshold as set as invalid.
weight_thershold (float): Removes samples below threshold weight. Only used if occupancy field is provided.
"""
super().__init__(
num_samples=num_samples,
spacing_fn=lambda x: 1 / x,
spacing_fn_inv=lambda x: 1 / x,
train_stratified=train_stratified,
occupancy_field=occupancy_field,
weight_threshold=weight_threshold,
)
num_samples = num_samples or self.num_samples
num_rays = ray_bundle.origins.shape[0]
bins = torch.linspace(0.0, 1.0, num_samples + 1).to(ray_bundle.origins.device) # shape (num_samples+1)
bins = ray_bundle.nears[:, None] + bins[None, :] * (
ray_bundle.fars[:, None] - ray_bundle.nears[:, None]
) # shape (num_rays, num_samples+1)
class SqrtSampler(SpacedSampler):
"""Square root sampler along a ray"""
if self.train_stratified and self.training:
t_rand = torch.rand((num_rays, num_samples + 1), dtype=bins.dtype, device=bins.device)
bin_centers = (bins[..., 1:] + bins[..., :-1]) / 2.0
bin_upper = torch.cat([bin_centers, bins[..., -1:]], -1)
bin_lower = torch.cat([bins[..., :1], bin_centers], -1)
bins = bin_lower + (bin_upper - bin_lower) * t_rand
def __init__(
self,
num_samples: int,
train_stratified=True,
occupancy_field: Optional[OccupancyGrid] = None,
weight_threshold: float = 1e-4,
) -> None:
"""
Args:
num_samples (int): Number of samples per ray
train_stratified (bool): Use stratified sampling during training. Defults to True
occupancy_field (OccupancyGrid, optional): Occupancy grid. If provides,
samples below weight_threshold as set as invalid.
weight_thershold (float): Removes samples below threshold weight. Only used if occupancy field is provided.
"""
super().__init__(
num_samples=num_samples,
spacing_fn=torch.sqrt,
spacing_fn_inv=lambda x: x**2,
train_stratified=train_stratified,
occupancy_field=occupancy_field,
weight_threshold=weight_threshold,
)
ray_samples = ray_bundle.get_ray_samples(bin_starts=bins[..., :-1], bin_ends=bins[..., 1:])
return ray_samples
class LogSampler(SpacedSampler):
"""Log sampler along a ray"""
def __init__(
self,
num_samples: int,
train_stratified=True,
occupancy_field: Optional[OccupancyGrid] = None,
weight_threshold: float = 1e-4,
) -> None:
"""
Args:
num_samples (int): Number of samples per ray
train_stratified (bool): Use stratified sampling during training. Defults to True
occupancy_field (OccupancyGrid, optional): Occupancy grid. If provides,
samples below weight_threshold as set as invalid.
weight_thershold (float): Removes samples below threshold weight. Only used if occupancy field is provided.
"""
super().__init__(
num_samples=num_samples,
spacing_fn=torch.log,
spacing_fn_inv=torch.exp,
train_stratified=train_stratified,
occupancy_field=occupancy_field,
weight_threshold=weight_threshold,
)
class PDFSampler(Sampler):
+58 -1
View File
@@ -2,7 +2,7 @@
Test samplers
"""
import torch
from pyrad.graphs.modules.ray_sampler import PDFSampler, UniformSampler
from pyrad.graphs.modules.ray_sampler import LinearDisparitySampler, LogSampler, PDFSampler, SqrtSampler, UniformSampler
from pyrad.cameras.rays import RayBundle
from pyrad.graphs.modules.scene_colliders import NearFarCollider
@@ -29,6 +29,63 @@ def test_uniform_sampler():
# TODO Tancik: Add more precise tests
def test_lin_disp_sampler():
"""Test linear in disparity sampler"""
near_plane = 2
far_plane = 4
num_samples = 15
sampler = LinearDisparitySampler(num_samples=num_samples)
origins = torch.zeros((10, 3))
directions = torch.ones_like(origins)
radius = torch.ones((10, 1))
ray_bundle = RayBundle(origins=origins, directions=directions, pixel_area=radius)
collider = NearFarCollider(near_plane, far_plane)
ray_bundle = collider(ray_bundle)
ray_samples = sampler(ray_bundle)
assert ray_samples.frustums.get_positions().shape[-2] == num_samples
def test_sqrt_sampler():
"""Test square root sampler"""
near_plane = 2
far_plane = 4
num_samples = 15
sampler = SqrtSampler(num_samples=num_samples)
origins = torch.zeros((10, 3))
directions = torch.ones_like(origins)
radius = torch.ones((10, 1))
ray_bundle = RayBundle(origins=origins, directions=directions, pixel_area=radius)
collider = NearFarCollider(near_plane, far_plane)
ray_bundle = collider(ray_bundle)
ray_samples = sampler(ray_bundle)
assert ray_samples.frustums.get_positions().shape[-2] == num_samples
def test_log_sampler():
"""Test log sampler"""
near_plane = 2
far_plane = 4
num_samples = 15
sampler = LogSampler(num_samples=num_samples)
origins = torch.zeros((10, 3))
directions = torch.ones_like(origins)
radius = torch.ones((10, 1))
ray_bundle = RayBundle(origins=origins, directions=directions, pixel_area=radius)
collider = NearFarCollider(near_plane, far_plane)
ray_bundle = collider(ray_bundle)
ray_samples = sampler(ray_bundle)
assert ray_samples.frustums.get_positions().shape[-2] == num_samples
def test_pdf_sampler():
"""Test pdf sampler"""
near_plane = 2