Cameron Coward


One of the popular applications for machine learning is object detection. Objection detection models can identify a wide variety of real-world items in standard photos or video feeds. Last year, we featured a traffic-counting device built using OpenDataCam running on an NVIDIA Jetson TX2. Now the software has been upgraded to version 2.0, and that introduces a handful of handy new features.

The primary advantage of OpenDataCam over other software is that it’s intended specifically for identifying and tracking moving objects. Most other software simply tells you what it sees in any given frame of a video feed. OpenDataCam is capable of understanding how objects move from one frame to the next. That’s particularly useful for applications like traffic counting, where each vehicle should only be counted once — even if it shows up in many frames of the video. Users can also specify a specific region of the video frame that should be analyzed, so unimportant portions of the frame can be ignored.

Version 2.0 of OpenDataCam introduces a few new features that should be very useful. Those include a new website, a new and more polished interface, and compatibility with the new NVIDIA Jetson Nano. The Jetson Nano is a single-board computer designed specifically for machine learning, and OpenDataCam 2.0 can take advantage of its powerful GPU to identify objects quickly. The most exciting new feature, however, is probably the API. This allows users to export important data directly from the software, so that it can be used as needed. The API should make it much easier for users to work with the data being gathered.



Source link