Abstract
Computer vision is currently experiencing success in various domains due to the harnessing of deep learning strategies. In the case of precision agriculture, computer vision is being investigated for detecting fruits from orchards. However, such strategies limit too-high complexity computation that is impossible to embed in an automated device. Nevertheless, most investigation of fruit detection is limited to a single fruit, resulting in the necessity of a one-to-many object detection system. This paper introduces a generic detection mechanism named FruitDet, designed to be prominent for detecting fruits. The FruitDet architecture is designed on the YOLO pipeline and achieves better performance in detecting fruits than any other detection model. The backbone of the detection model is implemented using DenseNet architecture. Further, the FruitDet is packed with newer concepts: attentive pooling, bottleneck spatial pyramid pooling, and blackout mechanism. The detection mechanism is benchmarked using five datasets, which combines a total of eight different fruit classes. The FruitDet architecture acquires better performance than any other recognized detection methods in fruit detection.
Highlights
Precision agriculture [1] implements the knowledge gained through data processed by machinery and software to deal with uncertainties of agricultural systems
Intersection over union (IOU) = 0.2 is selected through a grid search approach over all datasets and baseline models (YOLOv3, YOLOv4, MangoYOLO, and FruitDet)
This paper introduces a fruit detection model named FruitDet to recognize multiple fruits in a single pipeline
Summary
Precision agriculture [1] implements the knowledge gained through data processed by machinery and software to deal with uncertainties of agricultural systems. The uncertainties of agricultural systems are dealt with various approaches, such as weather data, field sensor data, vision data, and so forth. Precision agriculture is required to achieve sustainability to fulfill the requirements of the current population growth and effectively handle the food-to-land demand. Autonomous harvesting systems often require vision systems to pinpoint the targeted objects, plants, or crops. Developing an autonomous harvesting system requires solving various vision-related problems: (a) is the fruit seen or not, (b) pinpointing fruit location, (c) is the fruit ready to be picked or not, (d) estimating the fruit load, (e) diseased or not, and so on
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
