-
您好，非常感谢您的工作。我想问一些关于数据增强方面的问题。
1.我有注意到你在faster_rcnn_RoITrans_r50_fpn_1x_dota1_5_gap512_msTrainTest_rotationTrainTest.py 这个文件中提到了用ms(muti-scale)多尺度的训练方式， 但是在train_config中的img_scale=(1024, 1024)只有一个尺…
-
Hello! In the paper (Section 4, Subsection D) Data augmentation), you mention Data Augmentation with randomly sampled magnitudes.
Unfortunately, random magnitudes of the transformations introduces qu…
-
[Graph contrastive learning with augmentations](https://proceedings.neurips.cc/paper_files/paper/2020/hash/3fe230348e9a12c13120749e3f9fa4cd-Abstract.html)
```bib
@article{you2020graph,
title={Gra…
-
## 🚀 Feature
With #2568, fill values for arbitrary number of channels was added for `warp_affine`. It would be very convenient to adapt the functions `affine` and `rotate`, and the corresponding augm…
-
### Correction Scope
some records - e.g. it's seems to be wrong for a number of records, but right for others
### Metadata Record Identifier
_No response_
### metadata field
variableMeasured
###…
-
In the paper, you mentioned adding noise to the training data and perturbeding the tone. May I know where the code for this part is? I couldn't find them. Thank you
-
One of the core challenges of machine learning for airborne biodiversity observation is trying to generalize across sensors and acquisition conditions. Given low data sizes, data augmentations are cru…
bw4sz updated
3 months ago
-
### 현재 상황과 문제점
본 Task에서 데이터 증강은 Question과 Answer에 대한 증강이 필요하다.
Context와 Answer를 고정시키고 Question을 생성하는 Question Generation 방법으로 Question과 Answer Pair를 생성할 수 있다.
### 개선 제안 사항
1. 모델 학습을 통한 Question 생성…
-
transforms
- [x] GridDistortion
- [x] ShiftScaleRotate
- [x] Cutout
- [x] ElasticTransform
- [x] RandomPerspective
- [x] AugMix
-
By having Online Data Augmentation feature, the library can improve its training feature while suppressing memory consumpton.
For images,
- Affine or Homography transformation
- Color/ brightnes…