TFDSv4 carla_video_tracking upgrade. #1829
Closed
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
…)]] to represent each batch rather than returning Dict[ndarray: (batch, frames, coords)].
jprokos26
changed the title
jprokos26
commented
Dec 20, 2022
armory/datasets/preprocessing.py
Outdated
Comment on lines
233
to
246
| # if batch.dtype == np.object: | ||
| # for x in batch: | ||
| # check_shapes(x.shape, context.x_shape) | ||
| # assert x.dtype == context.input_type | ||
| # assert x.min() >= context.input_min | ||
| # assert x.max() <= context.input_max | ||
|
|
||
| # quantized_batch = np.zeros_like(batch, dtype=np.object) | ||
| # for i in range(len(batch)): | ||
| # quantized_batch[i] = ( | ||
| # batch[i].astype(context.output_type) / context.quantization | ||
| # ) | ||
| # batch = quantized_batch | ||
|
|
There was a problem hiding this comment.
Should the preprocessor handle pure python objects as done here?
armory/datasets/preprocessing.py
Outdated
Comment on lines
257
to
264
| # assert tf.math.reduce_min(batch) >= context.input_min | ||
| # assert tf.math.reduce_max(batch) <= context.input_max | ||
|
|
||
| # for x in batch: | ||
| # assert x.dtype == context.output_type | ||
| # assert tf.math.reduce_min(x) >= context.output_min | ||
| # assert tf.math.reduce_max(x) <= context.output_max | ||
|
|
There was a problem hiding this comment.
Since this is now passed into the dataloader as a function instead of operating on the actual tensor, should this still check the min & max values? TF was complaining about assert statements, but I can look for another way to implement this logic.
Comment on lines
+172
to
+179
| class EvalGenerator: | ||
| """ | ||
| Wraps a specified number of batches in a DataGenerator to allow for evaluating on | ||
| part of a dataset when running through a scenario | ||
| """ | ||
|
|
||
| def __init__(self, armory_generator, num_eval_batches): | ||
| raise NotImplementedError("EvalGenerator not implemented") |
There was a problem hiding this comment.
Added to satisfy import statement in another file I had not changed; should this be implemented or removed? I can find where the specific site where this needed to be imported if it is helpful.
Comment on lines
+241
to
+278
| class ClipFrames: | ||
| """ | ||
| Clip Video Frames | ||
| Assumes first two dims are (batch, frames, ...) | ||
| """ | ||
|
|
||
| def __init__(self, max_frames): | ||
| max_frames = int(max_frames) | ||
| if max_frames <= 0: | ||
| raise ValueError(f"max_frames {max_frames} must be > 0") | ||
| self.max_frames = max_frames | ||
|
|
||
| def __call__(self, batch): | ||
| return batch[:, : self.max_frames] | ||
|
|
||
|
|
||
| class ClipVideoTrackingLabels: | ||
| """ | ||
| Truncate labels for CARLA video tracking, when max_frames is set | ||
| """ | ||
|
|
||
| def __init__(self, max_frames): | ||
| max_frames = int(max_frames) | ||
| if max_frames <= 0: | ||
| raise ValueError(f"max_frames {max_frames} must be > 0") | ||
| self.max_frames = max_frames | ||
|
|
||
| def clip_boxes(self, boxes): | ||
| return boxes[:, : self.max_frames, :] | ||
|
|
||
| def clip_metadata(self, patch_metadata_dict): | ||
| return { | ||
| k: v[:, : self.max_frames, ::] for (k, v) in patch_metadata_dict.items() | ||
| } | ||
|
|
||
| def __call__(self, x, labels): | ||
| boxes, patch_metadata_dict = labels | ||
| return self.clip_boxes(boxes), self.clip_metadata(patch_metadata_dict) |
There was a problem hiding this comment.
Do these classes belong here or elsewhere?
|
|
||
|
|
||
| @register | ||
| def carla_video_tracking_dev(element, max_frames=None): |
There was a problem hiding this comment.
I am assuming max_frames gets passed in as a dataset_kwargs but this is untested as I cannot find an example of it being used previously.
… much about keeping track of module name, function name and how it registers within armory to global variable 'supported', a MetricNameSpace, and how it is passed as input to GlobalMeter/Meter creation in MetricsLogger
* update ci tests * combine dockerfiles * update image map * remove carla from `build.py` * fix issue 1787 * remove `tf2` container * remove `deepspeech` * update documentation
…)]] to represent each batch rather than returning Dict[ndarray: (batch, frames, coords)].
|
Replaced by #1836 |
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.
Throws an error when run without flags but completes successfully with
--skip-attack. See #1814 for more information.Status:
carla_video_tracking_devcarla_video_trackingBoth uploaded to s3 and ready for review.