Localizing Small Apples in Complex Apple Orchard Environments
The localization of fruits is an essential first step in automated agricultural pipelines for yield estimation or fruit picking. One example of this is the localization of apples in images of entire apple trees. Since the apples are very small objects in such scenarios, we tackle this problem by adapting the object proposal generation system AttentionMask that focuses on small objects. We adapt AttentionMask by either adding a new module for very small apples or integrating it into a tiling framework. Both approaches clearly outperform standard object proposal generation systems on the MinneApple dataset covering complex apple orchard environments. Our evaluation further analyses the improvement w.r.t. the apple sizes and shows the different characteristics of our two approaches.