代码拉取完成,页面将自动刷新
A pytorch reimplementation of DD-Net, which is a lightweight network for body/hand action recognition.
A subset of preprocessd JHMDB data has been in this repo, which is GT_test_1.pkl
,GT_train_1.pkl
.
The code is written and tested in Pytorch 1.3.0
.
git clone https://github.com/BlurryLight/DD-Net-Pytorch.git
python train.py --batch-size 512 --epochs 600 --lr 0.001 | tee train.log
The number of parameters is the same with the the original Keres version, which is 1.80M in JHMDB.
The val performance is comparable to the original version. On the provided JHMDB split, the performance is as blow.
此处可能存在不合适展示的内容,页面不予展示。您可通过相关编辑功能自查并修改。
如您确认内容无涉及 不当用语 / 纯广告导流 / 暴力 / 低俗色情 / 侵权 / 盗版 / 虚假 / 无价值内容或违法国家有关法律法规的内容,可点击提交进行申诉,我们将尽快为您处理。