Skip to content

A Lightweight Residual Graph CNN for Pedestrians Trajectory Prediction

Notifications You must be signed in to change notification settings

yanwu-ge/Res-GCNN

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

5 Commits
 
 
 
 
 
 

Repository files navigation

Res-GCNN

residual graph cnn for trajectory prediction, used in autonomous vehicles ETH HOTEL UNIV ZARA1 ZARA2 AVG Social-STGCNN 0.36/0.63 0.64/1.11 0.37/0.66 0.49/0.85 0.37/0.65 0.44/0.79 0.36/0.63 0.34/0.53 0.35/0.58 0.30/0.48 0.36/0.63 0.44/0.75 250+2txp+tcn3x1

  • all res 0.62/1.13 0.35/0.55 0.43/0.84 0.31/0.53 0.27/0.45 0.39/0.70 200epoch复现 3740paras 0.71/1.51 0.36/0.64 0.41/0.75 0.31/0.52 0.26/0.43 修高model.py184行,self.n_txpcnn-0 5.06/5.97 0.81/1.69 4.93/5.75 0.44/0.78 1.62/1.63 0.41/0.77 1.53/1.54 0.31/0.52 1.45/1.47 0.26/0.44 2.92/3.27

修高model.py184行,self.n_txpcnn-1 +188行 model_2 0.35/0.61 0.65/1.23 0.34/0.61 0.32/0.52 0.29/0.50 0.40/0.77 0.34/0.60 0.30/0.50 0.35/0.63 0.26/0.44 0.33/0.59 0.39/0.69 配置同上行,epoch由200改为250 Model4 0.34/0.61 0.67/1.33 0.34/0.61 0.34/0.52 0.29/0.49 0.39/0.75 0.33/0.59 0.29/0.50 0.34/0.62 0.25/0.42 0.33/0.58 0.39/0.71 配置同上行,epoch由200改为170 Model5 0.34/0.60 0.78/1.64 0.34/0.62 0.34/0.56 0.29/0.51 0.40/0.76 0.33/0.60 0.30/0.51 0.34/0.62 0.26/0.44 0.33/0.59 0.42/0.78 修高model.py184行,self.n_txpcnn-1 +187行 1.40/2.17 0.69/1.41 1.27/1.97 0.30/0.46 1.11/1.56 0.39/0.73 1.08/1.62 0.30/0.52 1.01/1.56 0.25/0.43 1.17/1.77 0.39/0.71 改为Adam,150epochs Model_7 0.32/0.58 0.85/1.39 0.32/0.58 0.35/0.53 0.30/0.49 0.39/0.73 0.33/0.60 0.29/0.52 0.35/0.62 0.24/0.41 0.33/0.58 0.42/0.72 打开Utils.pu 75 76行 Model_8

Releases

No releases published

Packages

No packages published