Nvidia Enhances Deep Learning SDK with Major Updates
With these major enhancements to deep learning SDK, Nvidia aims to improve GPU acceleration for computation intensive deep learning applications. First of all Nvidia has introduced object recognition with DIGITS4. This enables scientists and engineers to solve complex image recognition problems they faced earlier.
An application of this feature is vehicle tracking, something that can be used by organizations for data and planning regarding logistics and shipments. This can also work with facial recognition application as we have seen in movies. Things seem to be getting closer and closer to the future with each advancement in technology. The progress is slow yet steady.
This new feature can also be used in the self-driving cars application.
This will allow the car to differentiate between different objects on the road and allow developers to program different, specific commands according to nearby objects. So the car knows exactly what to do in specific conditions.
The self driving car can differentiate between a bike, bus and car with this new deep learning SDK enhancement. With this application the car can detect signs and know when to slow down and when it is okay to go a bit faster.
Another application of this deep learning SDK enhancement is medical diagnostics where data can be read off x-rays and other tests in order to give a result. The CUDA based cuDNN has also been updated to version 5.1 in order to improve performance on VGG and ResNet networks.
These are deep learning neural networks that consist of multiple layers of data extractions. This will give them an edge in finding solutions to advanced problems.
DIGITS 4 is available now for download for registered Nvidia developers. On the other hand cuDNN 5.1 will be coming out late August this year. We are looking forward to other applications of this deep learning SDK enhancement.