Belief Regulated Dual Propagation Nets for Learning Action Effects on Articulated Multi-Part Objects

Here is code and dataset for [BRDPN].

Ahmet E. Tekden, Aykut Erdem, Erkut Erdem, Mert Imre, M. Yunus Seker, Emre Ugur

---

Requirements:

Models are trained in Python 2.7.

Required Packages: numpy, matplotlib, keras,tensorflow-gpu, tqdm, scikit-learn

You can also use:

conda env create -f environment.yml

Required FFMPEG codec for manimation on matplotlib. This is required for creating the animations for effect predictions in matplotlib.

sudo apt install ffmpeg

---

Datasets needs to be downloaded. Instructions can be found in Data folder.

Pretrained models are available in Models folder.

---

[Website] [Paper-pdf] [Video] [Arxiv]