Import data resampling #36
Open
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
and other PRAGMA options for better paralle read/write performance
as we don't always have values for all joints and then want to default to 0.0
texhnolyze
force-pushed
the
feature/import-data-resampling
branch
2 times, most recently
from
d585d15
to
5c3ce5b
Compare
with different converters by data type and different resampling strategies: - game_states are converted with original rate - images are converted with a max sample rate, meaning they can have a lesser rate, but not more while not having consitent time deltas - joint_state and joint_command are resampled in sync with an interpolation using the last available value at any sample point The initial db data point at `t_0`, having a relative timestamp of `0.0` is set only when all synced data becomes available (joint_state, joint_command, later also imu). The last available game_state will also be saved at this time point `t_0`. Before this no data will be saved to the database.
texhnolyze
force-pushed
the
feature/import-data-resampling
branch
from
5c3ce5b
to
02db4ad
Compare
jaagut
reviewed
Nov 28, 2024
| converter = self.synced_data_converter | ||
| case "/DynamixelController/command": | ||
| last_messages_by_topic.joint_command = ros_msg | ||
| converter = self.synced_data_converter |
| from ddlitlab2024.dataset.imports.model_importer import InputData, Sample | ||
|
|
||
|
|
||
| class OriginalRateResampler: |
There was a problem hiding this comment.
Please create an interface for Resamplers
| if num_samples > 0: | ||
| return samples | ||
| else: | ||
| self.last_received_data = data |
There was a problem hiding this comment.
why do you set last_reveived_data only here?
Comment on lines
+37
to
+39
| self.last_sampled_data = self.last_received_data | ||
| self.last_sample_step_timestamp = self.last_sample_step_timestamp + self.sampling_step_in_seconds | ||
| samples.append(Sample(data=self.last_sampled_data, timestamp=self.last_sample_step_timestamp)) |
There was a problem hiding this comment.
Using class variables in Sample might be problematic if used by copy
| else: | ||
| self.last_received_data = data | ||
| return [ | ||
| Sample(data=self.last_sampled_data, timestamp=self.last_sample_step_timestamp, was_sampled_already=True) |
|
|
||
| @abstractmethod | ||
| def convert_to_model(self, data: InputData, relative_timestamp: float, recording: Recording) -> ModelData: | ||
| """_summary_ |
| models = ModelData() | ||
|
|
||
| for sample in self.resampler.resample(data, relative_timestamp): | ||
| if not sample.was_sampled_already: |
|
|
||
| if img_scaling_changed: | ||
| logger.error( | ||
| "The image sizes changed, during one recording! All images of a recording must have the same size." |
There was a problem hiding this comment.
Suggested change
| "The image sizes changed, during one recording! All images of a recording must have the same size." | |
| "The image sizes changed during one recording! All images of a recording must have the same size." |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Summary
Proposed changes
This PR adds resampling with different converters by data type and different resampling strategies:
The initial db data point at
t_0, having a relative timestamp of0.0is set only when all synced data becomes available (joint_state, joint_command, later also imu).The last available game_state will also be saved at this time point
t_0.Before this, no data will be saved to the database.
The
Converters currently still contain a mixture of bitbots/rosbag specific data access and conversion logic, which can be easily adjusted however by changing the fields ofInputDatato generic types to which we convert all of our different input data.I think doing this abstraction makes most sense once we see the actual difference in data.