visualize_reconstruction fixes

Summary: Various fixes to get visualize_reconstruction running, and an interactive test for it.

Reviewed By: kjchalup

Differential Revision: D39286691

fbshipit-source-id: 88735034cc01736b24735bcb024577e6ab7ed336

projects/implicitron_trainer/README.md

@@ -66,7 +66,7 @@ If you have a custom `experiment.py` script (as in the Option 2 above), replace
 To run training, pass a yaml config file, followed by a list of overridden arguments.
 For example, to train NeRF on the first skateboard sequence from CO3D dataset, you can run:
 ```shell
-dataset_args=data_source_args.dataset_map_provider_JsonIndexDatasetMapProvider_args
+dataset_args=data_source_ImplicitronDataSource_args.dataset_map_provider_JsonIndexDatasetMapProvider_args
 pytorch3d_implicitron_runner --config-path ./configs/ --config-name repro_singleseq_nerf \
     $dataset_args.dataset_root=<DATASET_ROOT> $dataset_args.category='skateboard' \
     $dataset_args.test_restrict_sequence_id=0 test_when_finished=True exp_dir=<CHECKPOINT_DIR>
@@ -87,7 +87,7 @@ To run evaluation on the latest checkpoint after (or during) training, simply ad
 
 E.g. for executing the evaluation on the NeRF skateboard sequence, you can run:
 ```shell
-dataset_args=data_source_args.dataset_map_provider_JsonIndexDatasetMapProvider_args
+dataset_args=data_source_ImplicitronDataSource_args.dataset_map_provider_JsonIndexDatasetMapProvider_args
 pytorch3d_implicitron_runner --config-path ./configs/ --config-name repro_singleseq_nerf \
     $dataset_args.dataset_root=<CO3D_DATASET_ROOT> $dataset_args.category='skateboard' \
     $dataset_args.test_restrict_sequence_id=0 exp_dir=<CHECKPOINT_DIR> eval_only=True

projects/implicitron_trainer/tests/test_experiment.py

@@ -13,16 +13,7 @@ from hydra import compose, initialize_config_dir
 from omegaconf import OmegaConf
 
 from .. import experiment
-from .utils import intercept_logs
+from .utils import interactive_testing_requested, intercept_logs
-
-
-def interactive_testing_requested() -> bool:
-    """
-    Certain tests are only useful when run interactively, and so are not regularly run.
-    These are activated by this funciton returning True, which the user requests by
-    setting the environment variable `PYTORCH3D_INTERACTIVE_TESTING` to 1.
-    """
-    return os.environ.get("PYTORCH3D_INTERACTIVE_TESTING", "") == "1"
 
 
 internal = os.environ.get("FB_TEST", False)

projects/implicitron_trainer/tests/test_visualize.py

@@ -0,0 +1,27 @@
+# Copyright (c) Meta Platforms, Inc. and affiliates.
+# All rights reserved.
+#
+# This source code is licensed under the BSD-style license found in the
+# LICENSE file in the root directory of this source tree.
+
+import os
+import unittest
+
+from .. import visualize_reconstruction
+from .utils import interactive_testing_requested
+
+internal = os.environ.get("FB_TEST", False)
+
+
+class TestVisualize(unittest.TestCase):
+    def test_from_defaults(self):
+        if not interactive_testing_requested():
+            return
+        checkpoint_dir = os.environ["exp_dir"]
+        argv = [
+            f"exp_dir={checkpoint_dir}",
+            "n_eval_cameras=40",
+            "render_size=[64,64]",
+            "video_size=[256,256]",
+        ]
+        visualize_reconstruction.main(argv)

projects/implicitron_trainer/tests/utils.py

@@ -6,6 +6,7 @@
 
 import contextlib
 import logging
+import os
 import re
 
 
@@ -28,3 +29,12 @@ def intercept_logs(logger_name: str, regexp: str):
         yield intercepted_messages
     finally:
         logger.removeFilter(interceptor)
+
+
+def interactive_testing_requested() -> bool:
+    """
+    Certain tests are only useful when run interactively, and so are not regularly run.
+    These are activated by this funciton returning True, which the user requests by
+    setting the environment variable `PYTORCH3D_INTERACTIVE_TESTING` to 1.
+    """
+    return os.environ.get("PYTORCH3D_INTERACTIVE_TESTING", "") == "1"

projects/implicitron_trainer/visualize_reconstruction.py

@@ -5,10 +5,11 @@
 # This source code is licensed under the BSD-style license found in the
 # LICENSE file in the root directory of this source tree.
 
-"""Script to visualize a previously trained model. Example call:
+"""
+Script to visualize a previously trained model. Example call:
 
-    projects/implicitron_trainer/visualize_reconstruction.py
+    pytorch3d_implicitron_visualizer \
-    exp_dir='./exps/checkpoint_dir' visdom_show_preds=True visdom_port=8097
+    exp_dir='./exps/checkpoint_dir' visdom_show_preds=True visdom_port=8097 \
     n_eval_cameras=40 render_size="[64,64]" video_size="[256,256]"
 """
 
@@ -18,9 +19,9 @@ from typing import Optional, Tuple
 
 import numpy as np
 import torch
-from omegaconf import OmegaConf
+from omegaconf import DictConfig, OmegaConf
-from pytorch3d.implicitron.models.visualization import render_flyaround
+from pytorch3d.implicitron.models.visualization.render_flyaround import render_flyaround
-from pytorch3d.implicitron.tools.configurable import get_default_args
+from pytorch3d.implicitron.tools.config import enable_get_default_args, get_default_args
 
 from .experiment import Experiment
 
@@ -38,7 +39,7 @@ def visualize_reconstruction(
     visdom_server: str = "http://127.0.0.1",
     visdom_port: int = 8097,
     visdom_env: Optional[str] = None,
-):
+) -> None:
     """
     Given an `exp_dir` containing a trained Implicitron model, generates videos consisting
     of renderes of sequences from the dataset used to train and evaluate the trained
@@ -76,22 +77,27 @@ def visualize_reconstruction(
     config = _get_config_from_experiment_directory(exp_dir)
     config.exp_dir = exp_dir
     # important so that the CO3D dataset gets loaded in full
-    dataset_args = (
+    data_source_args = config.data_source_ImplicitronDataSource_args
-        config.data_source_args.dataset_map_provider_JsonIndexDatasetMapProvider_args
+    if "dataset_map_provider_JsonIndexDatasetMapProvider_args" in data_source_args:
-    )
+        dataset_args = (
-    dataset_args.test_on_train = False
+            data_source_args.dataset_map_provider_JsonIndexDatasetMapProvider_args
+        )
+        dataset_args.test_on_train = False
+        if restrict_sequence_name is not None:
+            dataset_args.restrict_sequence_name = restrict_sequence_name
+
     # Set the rendering image size
     model_factory_args = config.model_factory_ImplicitronModelFactory_args
+    model_factory_args.force_resume = True
     model_args = model_factory_args.model_GenericModel_args
     model_args.render_image_width = render_size[0]
     model_args.render_image_height = render_size[1]
-    if restrict_sequence_name is not None:
-        dataset_args.restrict_sequence_name = restrict_sequence_name
 
     # Load the previously trained model
-    experiment = Experiment(config)
+    experiment = Experiment(**config)
-    model = experiment.model_factory(force_resume=True)
+    model = experiment.model_factory(exp_dir=exp_dir)
-    model.cuda()
+    device = torch.device("cuda")
+    model.to(device)
     model.eval()
 
     # Setup the dataset
@@ -101,6 +107,11 @@ def visualize_reconstruction(
     if dataset is None:
         raise ValueError(f"{split} dataset not provided")
 
+    if visdom_env is None:
+        visdom_env = (
+            "visualizer_" + config.training_loop_ImplicitronTrainingLoop_args.visdom_env
+        )
+
     # iterate over the sequences in the dataset
     for sequence_name in dataset.sequence_names():
         with torch.no_grad():
@@ -114,23 +125,26 @@ def visualize_reconstruction(
                 n_flyaround_poses=n_eval_cameras,
                 visdom_server=visdom_server,
                 visdom_port=visdom_port,
-                visdom_environment=f"visualizer_{config.visdom_env}"
+                visdom_environment=visdom_env,
-                if visdom_env is None
-                else visdom_env,
                 video_resize=video_size,
+                device=device,
             )
 
 
-def _get_config_from_experiment_directory(experiment_directory):
+enable_get_default_args(visualize_reconstruction)
+
+
+def _get_config_from_experiment_directory(experiment_directory) -> DictConfig:
     cfg_file = os.path.join(experiment_directory, "expconfig.yaml")
     config = OmegaConf.load(cfg_file)
+    # pyre-ignore[7]
     return config
 
 
-def main(argv):
+def main(argv) -> None:
     # automatically parses arguments of visualize_reconstruction
     cfg = OmegaConf.create(get_default_args(visualize_reconstruction))
-    cfg.update(OmegaConf.from_cli())
+    cfg.update(OmegaConf.from_cli(argv))
     with torch.no_grad():
         visualize_reconstruction(**cfg)
 

pytorch3d/implicitron/dataset/single_sequence_dataset.py

@@ -9,7 +9,7 @@
 # provide data for a single scene.
 
 from dataclasses import field
-from typing import Iterable, List, Optional
+from typing import Iterable, Iterator, List, Optional, Tuple
 
 import numpy as np
 import torch
@@ -46,6 +46,12 @@ class SingleSceneDataset(DatasetBase, Configurable):
     def __len__(self) -> int:
         return len(self.poses)
 
+    def sequence_frames_in_order(
+        self, seq_name: str
+    ) -> Iterator[Tuple[float, int, int]]:
+        for i in range(len(self)):
+            yield (0.0, i, i)
+
     def __getitem__(self, index) -> FrameData:
         if index >= len(self):
             raise IndexError(f"index {index} out of range {len(self)}")

pytorch3d/implicitron/models/visualization/render_flyaround.py

@@ -61,7 +61,7 @@ def render_flyaround(
         "depths_render",
         "_all_source_images",
     ),
-):
+) -> None:
     """
     Uses `model` to generate a video consisting of renders of a scene imaged from
     a camera flying around the scene. The scene is specified with the `dataset` object and
@@ -133,6 +133,7 @@ def render_flyaround(
     seq_idx = list(dataset.sequence_indices_in_order(sequence_name))
     train_data = _load_whole_dataset(dataset, seq_idx, num_workers=num_workers)
     assert all(train_data.sequence_name[0] == sn for sn in train_data.sequence_name)
+    # pyre-ignore[6]
     sequence_set_name = "train" if is_train_frame(train_data.frame_type)[0] else "test"
     logger.info(f"Sequence set = {sequence_set_name}.")
     train_cameras = train_data.camera
@@ -209,7 +210,7 @@ def render_flyaround(
 
 def _load_whole_dataset(
     dataset: torch.utils.data.Dataset, idx: Sequence[int], num_workers: int = 10
-):
+) -> FrameData:
     load_all_dataloader = torch.utils.data.DataLoader(
         torch.utils.data.Subset(dataset, idx),
         batch_size=len(idx),
@@ -220,7 +221,7 @@ def _load_whole_dataset(
     return next(iter(load_all_dataloader))
 
 
-def _images_from_preds(preds: Dict[str, Any]):
+def _images_from_preds(preds: Dict[str, Any]) -> Dict[str, torch.Tensor]:
     imout = {}
     for k in (
         "image_rgb",
@@ -253,7 +254,7 @@ def _images_from_preds(preds: Dict[str, Any]):
     return imout
 
 
-def _stack_images(ims: torch.Tensor, size: Optional[Tuple[int, int]]):
+def _stack_images(ims: torch.Tensor, size: Optional[Tuple[int, int]]) -> torch.Tensor:
     ba = ims.shape[0]
     H = int(np.ceil(np.sqrt(ba)))
     W = H
@@ -281,7 +282,7 @@ def _show_predictions(
     ),
     n_samples=10,
     one_image_width=200,
-):
+) -> None:
     """Given a list of predictions visualize them into a single image using visdom."""
     assert isinstance(preds, list)
 
@@ -329,7 +330,7 @@ def _generate_prediction_videos(
     video_path: str = "/tmp/video",
     video_frames_dir: Optional[str] = None,
     resize: Optional[Tuple[int, int]] = None,
-):
+) -> None:
     """Given a list of predictions create and visualize rotating videos of the
     objects using visdom.
     """
@@ -359,7 +360,7 @@ def _generate_prediction_videos(
             )
 
     for k in predicted_keys:
-        vws[k].get_video(quiet=True)
+        vws[k].get_video()
         logger.info(f"Generated {vws[k].out_path}.")
         if viz is not None:
             viz.video(

pytorch3d/implicitron/tools/video_writer.py

@@ -6,6 +6,7 @@
 
 import os
 import shutil
+import subprocess
 import tempfile
 import warnings
 from typing import Optional, Tuple, Union
@@ -15,6 +16,7 @@ import matplotlib.pyplot as plt
 import numpy as np
 from PIL import Image
 
+_DEFAULT_FFMPEG = os.environ.get("FFMPEG", "ffmpeg")
 
 matplotlib.use("Agg")
 
@@ -27,13 +29,13 @@ class VideoWriter:
     def __init__(
         self,
         cache_dir: Optional[str] = None,
-        ffmpeg_bin: str = "ffmpeg",
+        ffmpeg_bin: str = _DEFAULT_FFMPEG,
         out_path: str = "/tmp/video.mp4",
         fps: int = 20,
         output_format: str = "visdom",
         rmdir_allowed: bool = False,
         **kwargs,
-    ):
+    ) -> None:
         """
         Args:
             cache_dir: A directory for storing the video frames. If `None`,
@@ -74,7 +76,7 @@ class VideoWriter:
         self,
         frame: Union[matplotlib.figure.Figure, np.ndarray, Image.Image, str],
         resize: Optional[Union[float, Tuple[int, int]]] = None,
-    ):
+    ) -> None:
         """
         Write a frame to the video.
 
@@ -114,7 +116,7 @@ class VideoWriter:
         self.frames.append(outfile)
         self.frame_num += 1
 
-    def get_video(self, quiet: bool = True):
+    def get_video(self) -> str:
         """
         Generate the video from the written frames.
 
@@ -127,23 +129,39 @@ class VideoWriter:
 
         regexp = os.path.join(self.cache_dir, self.regexp)
 
-        if self.output_format == "visdom":  # works for ppt too
+        if shutil.which(self.ffmpeg_bin) is None:
-            ffmcmd_ = (
+            raise ValueError(
-                "%s -r %d -i %s -vcodec h264 -f mp4 \
+                f"Cannot find ffmpeg as `{self.ffmpeg_bin}`. "
-                       -y -crf 18 -b 2000k -pix_fmt yuv420p '%s'"
+                + "Please set FFMPEG in the environment or ffmpeg_bin on this class."
-                % (self.ffmpeg_bin, self.fps, regexp, self.out_path)
             )
+
+        if self.output_format == "visdom":  # works for ppt too
+            args = [
+                self.ffmpeg_bin,
+                "-r",
+                str(self.fps),
+                "-i",
+                regexp,
+                "-vcodec",
+                "h264",
+                "-f",
+                "mp4",
+                "-y",
+                "-crf",
+                "18",
+                "-b",
+                "2000k",
+                "-pix_fmt",
+                "yuv420p",
+                self.out_path,
+            ]
+
+            subprocess.check_call(args)
         else:
             raise ValueError("no such output type %s" % str(self.output_format))
 
-        if quiet:
-            ffmcmd_ += " > /dev/null 2>&1"
-        else:
-            print(ffmcmd_)
-        os.system(ffmcmd_)
-
         return self.out_path
 
-    def __del__(self):
+    def __del__(self) -> None:
         if self.tmp_dir is not None:
             self.tmp_dir.cleanup()