moritanian / cnnvr

WebVR with CNN depth prediction

CNNVR

WebVR with CNN depth prediction

Introduction

In the backend, CNN depth prediction create depth map from the web camera images. In the front end, it shows 3d movie.

Usage

environment (recommended)

python3.5 opencv-python 3.4 tensorflow-gpu 1.4

Get the trained model

As written in CNN depth prediction, download the TensorFlow trained model and place it in models/. If NYU_FCRN.ckpt does not exist and NYU_FCRN.ckpt.xxx exists, rename it to NYU_FCRN.ckpt.

Run the backend script

python predict.py ./models/NYU_FCRN.ckpt 

Front end

Run the web server and check the html/index.html!