[Feature] Support Charades Dataset #788
Conversation
Codecov ReportAttention: Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## 0.x #788 +/- ##
==========================================
+ Coverage 85.19% 85.24% +0.04%
==========================================
Files 130 132 +2
Lines 9418 9504 +86
Branches 1591 1626 +35
==========================================
+ Hits 8024 8102 +78
+ Misses 997 992 -5
- Partials 397 410 +13
Flags with carried forward coverage won't be shown. Click here to find out more. ☔ View full report in Codecov by Sentry. |
| data_prefix=None, | ||
| test_mode=False, | ||
| filename_tmpl_prefix='{}-', | ||
| filename_tmpl_suffix='{:06}.jpg', |
There was a problem hiding this comment.
I think it would be better to only have filename_tmpl rather than prefix & suffix:
- If you extract frames from the charades videos, it would be better to store frames of different videos in separate subfolders, which is a common practice.
- If the format is '{video_name}-{frame_idx}.jpg', you can also implement it with filename_tmpl. For example, you can use filename_tmpl.format(video_name, '{:06}') in the first step.
There was a problem hiding this comment.
cannot format {:06} with '{:06}', maybe the suffix is necessary.
Now, I format {} with '{:06}'
There was a problem hiding this comment.
That's OK, another possible solution is to use partial:
from functools import partial
tmpl = '{}_{:05d}.jpg'
tmpl_vid = partial(tmpl.format, 'blahblah')
frame_pth = tmpl_vid(10)
9f2b423 to
bf0f8f4
Compare
e6ff69f to
5231a6d
Compare
db3e6da to
0f77636
Compare
The preprocessing script hasn't been implemented yet. I use the slowfast annotations file here |
|
kindly ping :p @dreamerlin |
| ann_file_val = 'data/charades/annotations/charades_val_list_rawframes.csv' | ||
| ann_file_test = 'data/charades/annotations/charades_val_list_rawframes.csv' | ||
|
|
||
| mc_cfg = dict( |
| ann_file_val = 'data/charades/annotations/charades_val_list_rawframes.csv' | ||
| ann_file_test = 'data/charades/annotations/charades_val_list_rawframes.csv' | ||
|
|
||
| mc_cfg = dict( |
| @@ -125,40 +125,45 @@ def extract_feat(self, imgs): | |||
| x = self.backbone(imgs) | |||
| return x | |||
|
|
|||
| def average_clip(self, cls_score, num_segs=1): | |||
There was a problem hiding this comment.
Re-implement average_clip using aggregate_clip instead of removing it. And add a warning to tell user to use aggregate_clip. Also give the link of this pr for reference.
| def aggregate_clip(self, cls_score, num_segs=1): | ||
| if ('average_clips' not in self.test_cfg) and ('maximize_clips' | ||
| not in self.test_cfg): | ||
| raise KeyError( |
There was a problem hiding this comment.
Does this mean that all configs need to change?
| @@ -455,6 +455,50 @@ def __repr__(self): | |||
| return repr_str | |||
|
|
|||
|
|
|||
| @PIPELINES.register_module() | |||
| class SampleCharadesFrames(SampleFrames): | |||
There was a problem hiding this comment.
docstring missing.
Pls briefly explain the difference of this class and the existing ones to help users understand why Charades needs a spearate sampling
Uh oh!
There was an error while loading. Please reload this page.