search

seuqaj114 / paig

Code for the paper Physics-as-Inverse-Graphics: Joint Unsupervised Learning of Objects and Physics from Video
MIT License
38 stars 11 forks source link

Problem with dataset #4

Closed TsinghuaPhantom closed 4 years ago

TsinghuaPhantom commented 4 years ago

Hi, nice work, i have checked you dataset in Google drive and got confused. For example the data Datasets/3bp_color/color_3bp_vx2_vy2_sl40_r2_g60_m1_dt05.npz. The size of train_x is <class 'tuple'>: (10, 40, 36, 36, 3) which means the training set contains only 10 samples. Shouldn't is have more like 5000?

janacht commented 4 years ago

I'm having the same problem, what was your solution?

TsinghuaPhantom commented 4 years ago

I'm having the same problem, what was your solution?

Just generate the dataset by yourself using the code in https://github.com/seuqaj114/paig/blob/master/nn/datasets/generators.py.