-
Notifications
You must be signed in to change notification settings - Fork 9.2k
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
About multiscale training #78
Comments
You only need to modify the data = dict(
imgs_per_gpu=2,
workers_per_gpu=2,
train=dict(
type=dataset_type,
ann_file=data_root + 'annotations/instances_train2017.json',
img_prefix=data_root + 'train2017/',
img_scale=[(1333, 800), (1666, 1000)],
img_norm_cfg=img_norm_cfg,
size_divisor=32,
flip_ratio=0.5,
with_mask=True,
with_crowd=True,
with_label=True),
val=dict(
type=dataset_type,
ann_file=data_root + 'annotations/instances_val2017.json',
img_prefix=data_root + 'val2017/',
img_scale=(1333, 800),
img_norm_cfg=img_norm_cfg,
size_divisor=32,
flip_ratio=0,
with_mask=True,
with_crowd=True,
with_label=True),
test=dict(
type=dataset_type,
ann_file=data_root + 'annotations/instances_val2017.json',
img_prefix=data_root + 'val2017/',
img_scale=(1333, 800),
img_norm_cfg=img_norm_cfg,
size_divisor=32,
flip_ratio=0,
with_mask=False,
with_label=False,
test_mode=True)) |
Thanks, but my question is "what does multiscale training means". |
The term of "multiscale training" is adopted in many papers, which indicates resizing images to different scales at each iteration. In the challenge, we use the setting |
Thanks. |
…ated_dataset concatenated dataset fix
* [Tutorial Branch PR] add yolo detector for demo (open-mmlab#72) * patch * version * rm unsafe use * rm unsafe use * feat: obejct detection, init * [demo] add yolo detector * run successfully * [fix] root path, add necessary files * [fix] remove deprecated files, and * [doc] object detection, part 1 * [Doc] object detection, part 2 * [fix] remove unnecessary lines * [demo] change demo image * [fix] dependent module, name of tiny_motorbike * Trigger CI * [doc] fix ag.Categorical * Trigger CI * modify Jenkin * remove pandas * [docs] add cards, fix typos * [Jenkin] change back * [ name] classifier -> detector * [feat] add tqdm * [tqdm] show bar * [docs] add index.rst for object_detection * [docs] correct link to beginner.html, moving detection cards * torch tutorial (open-mmlab#71) * [Tutorial Branch PR] Tutorial patch (open-mmlab#73) * patch * version * rm unsafe use * rm unsafe use * fix rl searcher update * dataset wip * macos dataloader without using mx-opencv * data sampler and misc * rm * [Tutorial Branch PR] Modifications on docs (open-mmlab#78) * docs: retouch the tutorial of object detection * docs: correct links to tutorials of image classification * Trigger CI * [Tutorial Branch PR] Patch-1 (open-mmlab#77) * tutorial content for wednesday * logger info * patch+ * add files * tutorial updates * patch +1 * misc doc improvement * reduce time * ci * comments * warning and lazy import * lightgbm exception handling for macos * cut pytorch nas for now * warnings and miscs * te * split seed
I found that the repo has already supported using different image scales in the training.
This should be included in data augmentation, then what does multiscale training mean?
I think multiscale training should has similar format to input image pyramid, which has not been supported by this repo yet?
The text was updated successfully, but these errors were encountered: