Yield prediction is a significant factor in optimizing orchard management and improves fruit quality in woody crops. It helps farmers to make harvest and market plans, labor management, and provides vital information about orchard variability necessary for pruning, planning nutrients, and water management in future seasons. The most common approach for yield prediction in real practice consists of manually counting the fruit in a sample of trees, and multiplying the average number of fruits by the total number of trees in the orchard.
The traditional method for yield prediction in woody crops have some inconvenience: 1) the number of fruits on each tree uses to be variable, affecting the extrapolation to the whole orchard; 2) the sampled trees must be representative of the orchard; 3) final yield is not only affected by the number of fruits, but also by their size; 4) orchards can be installed on steep terrains that make difficult to perform the manual scouting; 5) overestimations could be made as a consequence of low productivity or missing plants.
Remote sensing has demonstrated its usefulness in agriculture allowing to information to be generated from whole orchards avoiding the limits of manual scouting in a wide range of applications, from fertilizer prescription to weed detection, plant phenotyping or water management. Among the remote sensing platforms one of the best suited for agriculture are Unmanned Aerial Vehicles (UAVs) because of their relatively low cost, the possibility of carrying a wide range of sensors, the ability to fly under the clouds, and the high resolution of the information acquired by their sensors.
In recent years, the advances in artificial intelligence (AI) and machine learning have made great achievements in object detection in images and 3D models. The application of such advances to images and 3D models generated from RGB cameras, multispectral sensors, and Lidar sensors on board ground platforms have proved to be useful in fruit detection and yield prediction in woody crops. However, ground platforms are usually too expensive and they are not as efficient as aerial platforms for acquiring information from whole orchards. Consequently, applying artificial intelligence and other analysis techniques to information acquired from sensors on board UAVs has potential in fruit detection and yield prediction, as some pioneering works have demonstrated.
We invite researchers to submit original research articles, review articles, and papers on novel methods in analyzing data acquired from UAVs concerning fruit detection and yield prediction in woody crops.
Contributions will cover, but are not limited to, the following:
• Application of AI techniques to fruit detection.
• Fruit detection from images or 3D models created from sensors on board UAVs.
• Yield prediction models from information acquired through UAV sensors.
• Definition of protocols for optimization of UAV and sensor configuration concerning fruit detection and yield prediction.
• Influence of alternate bearing on yield prediction models.
• Critical evaluation of main limits of UAV approach (shadows, fruit occlusion, light reflection, fruit overlapping, etc…).
Yield prediction is a significant factor in optimizing orchard management and improves fruit quality in woody crops. It helps farmers to make harvest and market plans, labor management, and provides vital information about orchard variability necessary for pruning, planning nutrients, and water management in future seasons. The most common approach for yield prediction in real practice consists of manually counting the fruit in a sample of trees, and multiplying the average number of fruits by the total number of trees in the orchard.
The traditional method for yield prediction in woody crops have some inconvenience: 1) the number of fruits on each tree uses to be variable, affecting the extrapolation to the whole orchard; 2) the sampled trees must be representative of the orchard; 3) final yield is not only affected by the number of fruits, but also by their size; 4) orchards can be installed on steep terrains that make difficult to perform the manual scouting; 5) overestimations could be made as a consequence of low productivity or missing plants.
Remote sensing has demonstrated its usefulness in agriculture allowing to information to be generated from whole orchards avoiding the limits of manual scouting in a wide range of applications, from fertilizer prescription to weed detection, plant phenotyping or water management. Among the remote sensing platforms one of the best suited for agriculture are Unmanned Aerial Vehicles (UAVs) because of their relatively low cost, the possibility of carrying a wide range of sensors, the ability to fly under the clouds, and the high resolution of the information acquired by their sensors.
In recent years, the advances in artificial intelligence (AI) and machine learning have made great achievements in object detection in images and 3D models. The application of such advances to images and 3D models generated from RGB cameras, multispectral sensors, and Lidar sensors on board ground platforms have proved to be useful in fruit detection and yield prediction in woody crops. However, ground platforms are usually too expensive and they are not as efficient as aerial platforms for acquiring information from whole orchards. Consequently, applying artificial intelligence and other analysis techniques to information acquired from sensors on board UAVs has potential in fruit detection and yield prediction, as some pioneering works have demonstrated.
We invite researchers to submit original research articles, review articles, and papers on novel methods in analyzing data acquired from UAVs concerning fruit detection and yield prediction in woody crops.
Contributions will cover, but are not limited to, the following:
• Application of AI techniques to fruit detection.
• Fruit detection from images or 3D models created from sensors on board UAVs.
• Yield prediction models from information acquired through UAV sensors.
• Definition of protocols for optimization of UAV and sensor configuration concerning fruit detection and yield prediction.
• Influence of alternate bearing on yield prediction models.
• Critical evaluation of main limits of UAV approach (shadows, fruit occlusion, light reflection, fruit overlapping, etc…).
