VISTA is now measuring at DC Visser Duiven
Thanks to the good network of Bricklog, a cooperation with Visser Duiven allows the VISTA partnership to capture full size video material in real-life. This started using a simple webcam to quickly gather initial footage to train the camera-based localization algorithms. As these algorithms are now based on neural networks that need to be trained, real-life camera images are important in order to provide the type of training data that will result in robust RTLS (Real Time Localization System) performance. The image gives a hint of algorithm performance using a simple (mono)webcam, looking through badly reflecting windows. Look at the rectangles that show the inferred positions of truck and trailer.