feat(t4 devkit): add support of saving rendering result as .rrd

[ktro2828]

Description

After this PR, users can save rendering result as follows:

from t4_devkit import Tier4

t4 = Tier4("annotation", "<DATASET_PATH>")

# Rendering scene 
scene_token = t4.scene[0].token
t4.render_scene(scene_token, save_dir="<DIR_TO_SAVE>")
# -> Rendering result will be saved to `save_dir/<AppID>.rrd`

Also, we can suspend to spawn a viewer by specifying show=False as below:

t4.render_scene(scene_token, save_dir="<DIR_TO_SAVE>", show=False)

Note that, both save_dir and show are keyword-only arguments.

How to review

Check whether each rendering method works including render_scene(...)/render_instance(...)/render_point_cloud(...).

How to test

test data

TIER IV INTERNAL LINK

test command

from t4_devkit import Tier4

t4 = Tier4("annotation", "<DATASET_PATH>")

# Rendering scene 
scene_token = t4.scene[0].token
t4.render_scene(scene_token, save_dir="<DIR_TO_SAVE>")
# -> Rendering result will be saved to `save_dir/<AppID>.rrd`

# Other rendering methods
# ...

Reference

Notes for reviewer

This PR fixes rerun=0.17.0 because some breaking changes are contained in the release of v0.18.0: https://github.com/rerun-io/rerun/releases/tag/0.18.0.

Edited on Aug 20
Please update your dependency first as follows:

poetry update t4-devkit

[kminoda]

@ktro2828
I got this error when running

from t4_devkit import Tier4
path = "/home/minoda/ghq/github.com/tier4/autoware-ml/data/t4dataset/database_v1_3/2ea987cf-2900-4ea6-a6e5-d71290b13601/0/"
t4 = Tier4("annotation", path)
scene_token = t4.scene[0].token
t4.render_scene(scene_token, "/home/minoda/Downloads/hogehoge")

Which is

======
Loading T4 tables in `annotation`...
Reverse indexing...
Done reverse indexing in 0.002 seconds.
======
9 attribute
7 calibrated_sensor
6 category
1373 ego_pose
42 instance
1 log
1 map
20 sample
598 sample_annotation
1337 sample_data
4 visibility
7 sensor
1 scene
0 object_ann
0 surface_ann
0 keypoint
Done loading in 0.023 seconds.
======
[2024-08-20T06:01:48Z INFO  re_sdk::spawn] A process is already listening at this address. Assuming it's a Rerun Viewer. addr=0.0.0.0:9876
[2024-08-20T06:01:48Z INFO  re_sdk_comms::server] New SDK client connected: 127.0.0.1:50900
[2024-08-20T06:01:48Z INFO  re_sdk_comms::server] New SDK client connected: 127.0.0.1:50908
Traceback (most recent call last):
  File "/home/minoda/ghq/github.com/tier4/tier4_perception_dataset/tmp.py", line 10, in <module>
    t4.render_scene(scene_token, "/home/minoda/Downloads/hogehoge")
  File "/home/minoda/.local/lib/python3.10/site-packages/t4_devkit/tier4.py", line 688, in render_scene
    max_timestamp_us = first_lidar_sd_record.timestamp + sec2us(max_time_seconds)
  File "/home/minoda/.local/lib/python3.10/site-packages/t4_devkit/common/timestamp.py", line 29, in sec2us
    return 1e6 * timestamp
TypeError: can't multiply sequence by non-int of type 'float'

Still haven't investigated in depth, but did you encounter a similar issue?

[ktro2828]

@ktro2828 I got this error when running

from t4_devkit import Tier4
path = "/home/minoda/ghq/github.com/tier4/autoware-ml/data/t4dataset/database_v1_3/2ea987cf-2900-4ea6-a6e5-d71290b13601/0/"
t4 = Tier4("annotation", path)
scene_token = t4.scene[0].token
t4.render_scene(scene_token, "/home/minoda/Downloads/hogehoge")

Which is

======
Loading T4 tables in `annotation`...
Reverse indexing...
Done reverse indexing in 0.002 seconds.
======
9 attribute
7 calibrated_sensor
6 category
1373 ego_pose
42 instance
1 log
1 map
20 sample
598 sample_annotation
1337 sample_data
4 visibility
7 sensor
1 scene
0 object_ann
0 surface_ann
0 keypoint
Done loading in 0.023 seconds.
======
[2024-08-20T06:01:48Z INFO  re_sdk::spawn] A process is already listening at this address. Assuming it's a Rerun Viewer. addr=0.0.0.0:9876
[2024-08-20T06:01:48Z INFO  re_sdk_comms::server] New SDK client connected: 127.0.0.1:50900
[2024-08-20T06:01:48Z INFO  re_sdk_comms::server] New SDK client connected: 127.0.0.1:50908
Traceback (most recent call last):
  File "/home/minoda/ghq/github.com/tier4/tier4_perception_dataset/tmp.py", line 10, in <module>
    t4.render_scene(scene_token, "/home/minoda/Downloads/hogehoge")
  File "/home/minoda/.local/lib/python3.10/site-packages/t4_devkit/tier4.py", line 688, in render_scene
    max_timestamp_us = first_lidar_sd_record.timestamp + sec2us(max_time_seconds)
  File "/home/minoda/.local/lib/python3.10/site-packages/t4_devkit/common/timestamp.py", line 29, in sec2us
    return 1e6 * timestamp
TypeError: can't multiply sequence by non-int of type 'float'

Still haven't investigated in depth, but did you encounter a similar issue?

@kminoda It seems to be caused by wrong usage of the method. render_scene expects following parameters:

def render_scene(
        self,
        scene_token: str,
        *,
        max_time_seconds: float = np.inf,
        save_dir: str | None = None,
        show: bool = True,
    ) -> None:

Then, you are specifying max_time_seconds="/home/minoda/Downloads/hogehoge".

However, from this PR, parameters below max_time_seconds... were key word-only arguments.

I'm sorry I've not noted that you need to update dependency first with poetry.
Could you try to run poetry update t4-devkit first?

After that, please try out:

...
t4.render_scene(<scene_token>, save_dir=<your_save_dir>)

[kminoda]

LGTM (sorry for not being able to review this for a while 🙏 )