[CodeCamp #1503] Add InstanceSegMetric to MMEval #70
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
|
|
ice-tong
reviewed
Jan 2, 2023
ice-tong
reviewed
Jan 4, 2023
| - pts_semantic_mask(numpy.ndarray): Ground truth semantic masks. | ||
| """ | ||
| for prediction, groundtruth in zip(predictions, groundtruths): | ||
| self._results.append((deepcopy(prediction), deepcopy(groundtruth))) |
There was a problem hiding this comment.
If we don't use deepcopy, the metric itself will change the value of the tensor inplace which may cause confusion.
Co-authored-by: yancong <32220263+ice-tong@users.noreply.github.com>
Co-authored-by: yancong <32220263+ice-tong@users.noreply.github.com>
ice-tong
reviewed
Jan 17, 2023
mmeval/metrics/instance_seg.py
Outdated
Comment on lines
161
to
172
| gt_semantic_masks = [] | ||
| gt_instance_masks = [] | ||
| pred_instance_masks = [] | ||
| pred_instance_labels = [] | ||
| pred_instance_scores = [] | ||
|
|
||
| for result_pred, result_gt in results: | ||
| gt_semantic_masks.append(result_gt['pts_semantic_mask']) | ||
| gt_instance_masks.append(result_gt['pts_instance_mask']) | ||
| pred_instance_masks.append(result_pred['pts_instance_mask']) | ||
| pred_instance_labels.append(result_pred['instance_labels']) | ||
| pred_instance_scores.append(result_pred['instance_scores']) |
There was a problem hiding this comment.
A suggestion that simplify the code~
Suggested change
| gt_semantic_masks = [] | |
| gt_instance_masks = [] | |
| pred_instance_masks = [] | |
| pred_instance_labels = [] | |
| pred_instance_scores = [] | |
| for result_pred, result_gt in results: | |
| gt_semantic_masks.append(result_gt['pts_semantic_mask']) | |
| gt_instance_masks.append(result_gt['pts_instance_mask']) | |
| pred_instance_masks.append(result_pred['pts_instance_mask']) | |
| pred_instance_labels.append(result_pred['instance_labels']) | |
| pred_instance_scores.append(result_pred['instance_scores']) | |
| gt_semantic_masks = [gt['pts_semantic_mask'] for _, gt in results] | |
| gt_instance_masks = [gt['pts_instance_mask'] for _, gt in results] | |
| pred_instance_masks = [pred['pts_instance_mask'] for pred, _ in results] | |
| pred_instance_labels = [pred['instance_labels'] for pred, _ in results] | |
| pred_instance_scores = [pred['instance_scores'] for pred, _ in results] |
There was a problem hiding this comment.
In this way, we would do 5 for loops, which may be time consuming.
mmeval/metrics/instance_seg.py
Outdated
Comment on lines
21
to
32
| Example: | ||
| >>> import numpy as np | ||
| >>> from mmeval import InstanceSegMetric | ||
| >>> seg_valid_class_ids = (3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 14, 16, 24, | ||
| >>> 28, 33, 34, 36, 39) | ||
| >>> class_labels = ('cabinet', 'bed', 'chair', 'sofa', 'table', 'door', | ||
| ... 'window', 'bookshelf', 'picture', 'counter', 'desk', | ||
| ... 'curtain', 'refrigerator', 'showercurtrain', 'toilet', | ||
| ... 'sink', 'bathtub', 'garbagebin') | ||
| >>> dataset_meta = dict( | ||
| ... seg_valid_class_ids=seg_valid_class_ids, classes=class_labels) | ||
| >>> |
There was a problem hiding this comment.
The example is too complicated and needs to be simplified~
There was a problem hiding this comment.
It is hard to simplify it for this metric need many code to generate input.
mmeval/metrics/instance_seg.py
Outdated
| infos.append(info) | ||
| return infos | ||
|
|
||
| def rename_gt(self, gt_semantic_masks, gt_instance_masks, valid_class_ids): |
There was a problem hiding this comment.
The rename in function name rename_gt seems a bit weird, what is be renamed?
There was a problem hiding this comment.
It seems that this function rename the input ground truth in order to evaluate in ScanNet protocol provided by ScanNet repo.
Co-authored-by: yancong <32220263+ice-tong@users.noreply.github.com>
Co-authored-by: yancong <32220263+ice-tong@users.noreply.github.com>
Co-authored-by: yancong <32220263+ice-tong@users.noreply.github.com>
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.
Motivation
Add InstanceSegMetric to MMEval
pr in mmdetection3d: openmmlab/mmdetection3d
Modification