Wind and drag turn out to be the biggest enemies for professional or casual bicyclists. An aerodynamic technique called slipstreaming (drafting), allows riders to take advantage of the airflow around fast-moving objects in an effort to reduce drag which helps them save as much as one-third of the energy consumed otherwise.
Slixstream in its current form is an Android App for devices running Android Oreo and above. The Android app detects a class of fast moving objects aka 'slipstreamable' objects in real time, identifies the trailing space behind those objects which they could slipstream into and draws bounding boxes around them. It also draws an alert when it identifies commonly tagged road signs.
The change in paradigm in hardware architecture to bring inference engines on to the device instead of offloading them to the cloud offers opportunities to bring compute-intensive computer vision applications to the mobile devices . We believe this technology could assist the riders to help squeeze out the last drop of performance and given the weight and power constraints for bicyclists, deep learning could bring about a great synergy between physical constraints and the power of mobile computing.
No need to send a request over a network connection and wait for a response. Critical for video applications that process successive frames coming from a camera.
New hardware specific to neural networks processing provide significantly faster computation than with general-use CPU alone.
The application runs even when outside of network coverage. You are not alone even on the most remote trails.
You are the king/queen of your data. The data does not leave the device.