Solution for remote real-time visual expertise of agricultural objects
¹University of Life Sciences and Technologies, Faculty of Information Technologies, Department of Computer Systems, Liela iela 2, LV-3001 Jelgava, Latvia
²University of Kassel, Process and Systems Engineering in Agriculture Group, Department of Agricultural and Biosystems Engineering, Nordbahnhofstrasse 1a, D-37213, Witzenhausen, Germany.
*Correspondence: aleksejs.zacepins@llu.lv
Abstract:
In recent years automated image and video analyses of plants and animals have become important techniques in Precision Agriculture for the detection of anomalies in development. Unlikely, machine learning (i.e., artificial neural networks, support vector machine, and other relevant techniques) are not always able to support decision making. Nevertheless, experts can use these techniques for developing more precise solutions and analysis approaches. It is labour-intensive and time-consuming for the experts to continuously visit the production sites to make direct on-site observations. Therefore, videos from the site need to be made available for remote viewing and analysis. In some cases it is also essential to monitor different parts of objects in agriculture and animal farming (e.g., bottom of the plants, stomach of the animal, etc.) which are difficult to access in standard recording procedures. One possible solution for the farmer is the use of a portable camera with real-streaming option rather than a stationary camera.
The aim of this paper is the proposition of a solution for real-time video streaming of agricultural objects (plants and/or animals) for remote expert evaluation and diagnosis. The proposed system is based on a Raspberry Pi 3, which is used to transfer the video from the attached camera to the YouTube streaming service. Users will be able to watch the video stream from the YouTube service on any device that has a web browser. Several cameras (USB, and Raspberry Pi camera) and video resolutions (from 480p till 1,080p) are compared and analysed, to find the best option, taking into account video quality, frame rates, and latency. Energy consumption of the whole system is evaluated and for the chosen solution it is 645 mA.
Key words:
precision agriculture, video streaming, visual observations