-
Notifications
You must be signed in to change notification settings - Fork 1.2k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[Feature] ACSampler #381
[Feature] ACSampler #381
Conversation
Codecov Report
@@ Coverage Diff @@
## master #381 +/- ##
==========================================
- Coverage 84.65% 84.31% -0.35%
==========================================
Files 121 102 -19
Lines 8507 7127 -1380
Branches 1394 1157 -237
==========================================
- Hits 7202 6009 -1193
+ Misses 952 898 -54
+ Partials 353 220 -133
Flags with carried forward coverage won't be shown. Click here to find out more.
Continue to review full report at Codecov.
|
frame_dir = line_split[idx] | ||
if self.audio_prefix is not None: | ||
audio_path = osp.join(self.audio_prefix, | ||
frame_dir) + '.npy' |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
audio_path = osp.join(self.audio_prefix, frame_dir + '.npy')
frame_dir = line_split[idx] | ||
if self.audio_prefix is not None: | ||
audio_path = osp.join(self.audio_prefix, | ||
frame_dir) + '.npy' |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
audio_path = osp.join(self.audio_prefix, frame_dir+'.npy')
frame_dir) + '.npy' | ||
if self.video_prefix: | ||
video_path = osp.join(self.video_prefix, | ||
frame_dir) + '.mp4' |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Same
img, (new_w, new_h), interpolation=self.interpolation) | ||
for img in results['imgs'] | ||
] | ||
import pdb |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
remove it
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
OK, the PR is still in progress
# import pdb | ||
# pdb.set_trace() | ||
test_crop = self.test_cfg.get('test_crop', 3) | ||
test_clip = self.test_cfg.get('test_clip', 30) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Can this two var be inferred by batch_size
?
assert (0 <= self.k_prime < self.top_k) and isinstance( | ||
self.k_prime, int) | ||
else: | ||
raise NotImplementedError |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
combine_type = combination.get('type', None)
assert combine_type in ('convex', 'av_union_list')
if combine_type == ...:
...
elif combine_type == ...:
...
Minor.
Minor. Minor. To squash. Minor.
This PR adds ACSampler from the paper: SCSampler, including:
This PR is challenging, as it is the first model that has a complicated pipeline operated on four kinds of modalities.
ACSampler is the 'simplified' version introduced in the paper, that trains a sampler offline and can be adapted to all kinds of 3D models. As for 'SCSampler'(which requires joint training the recognizer along with the sampler), it will be addressed after ACSampler is re-implemented.