-
## 一言でいうと
DNNの判断根拠を理解するための試みで、ある学習データ(サンプル)がなかった場合のモデルへの影響を手がかりにする。通常だと該当サンプルを抜いての再学習が必要だが、該当サンプルのlossを増減させた場合の最適解を既存の最適解から導出するという技を使いこれをクリアしている。
### 論文リンク
https://arxiv.org/abs/1703.04730
…
-
Hi,
I am training the yolov4-p5.cfg with the COCO dataset at resolution 896x896. I am using this command:
darknet detector train cfg/coco.data cfg/yolov4-p5.cfg yolov4-p5.conv.232
The weights…
-
The feature maps have dimensions of eccentricity and angle, rather than horizontal and vertical positions. So we do not have translation equivariance. Also, the image statistics may be different in th…
-
Does anyone have any background on why the values of the 3 parameters of the adam optimizer were chosen? In particular for the learning rate, too low initial learning rates can produce poor accuracy w…
-
Sorry, I didn't see the Expression loss (eq 10 in the paper) implemented in the training codes.
-
## 一言でいうと
従来のCNNのような複雑なアーキテクチャを利用せず，CNNの基本的な畳み込み特徴変換プロセスを、モデルアーキテクチャをチューニングせずに改善することを検討
新しい自己校正畳み込みを提示する
### 論文リンク
http://openaccess.thecvf.com/content_CVPR_2020/papers/Liu_Improving_Convolutional…
-
The citation [28] should be [27] in your experiments section?
-
I just released a new version of OpenAccess_EPUB (0.5.1) on PyPI, and it comes with a lot of new changes. One small change is that I added a wheel distribution (though lxml is still _by far_ the bigg…
-
Submission name: Things We've Learned at OpenCon
Contact lead: @Daniel-Mietchen
Issue area: #OpenAccess, #OpenData, #OpenEducation, #OpenResearch
Region: #Global
Issue Type: #Project
…
-
https://openaccess.thecvf.com/content_CVPR_2020/papers/Yao_BlendedMVS_A_Large-Scale_Dataset_for_Generalized_Multi-View_Stereo_Networks_CVPR_2020_paper.pdf
BlendedMVS: A Large-scale Dataset for Gene…