convolutional Patch Networks with Spatial Prior for Road Detection and Urban Scene understanding

convolutional Patch Networks with Spatial Prior for Road Detection and Urban Scene understanding

Deep learning ideas more and more fire, in this year's CVPR 2015 articles related articles have more than 20, can be seen is very fire. I am also happy to see this article in the recent context of semantic cutting and scene parsing.

ausingaofandandwhileattheof Jena.

The article says it can perform well on Windows, Mac and Linux systems, and can support OpenCL, Intel MKL, AMD ACML for Parallel computing acceleration. It looks really good!

Official Source code Address:
[CN24]

Let's start with the experimental results:
Experimental environment [Ubuntu 15.04]. Temporary no practical GPU acceleration

Code Programs folder:

Results after execution:
Single picture test results, the left side is the result of semantic cutting, the right side is the real measurement attempt.

No practical GPU acceleration, it is very slow to deal with a picture, time-consuming or quite a lot, for example, as seen in:

The following is the groping code. The next section will continue to share the learning process, thank you!

convolutional Patch Networks with Spatial Prior for Road Detection and Urban Scene understanding