Finished! Looks like this project is out of data at the moment!
Research
Why cocoa?
Theobroma cacao L. is the tropical tree producing cacao beans, the raw material for the chocolate industry as well as for many other products derived from the beans.
However, the global demand of cacao (Theobroma cacao L.) is expected to reach between 4.7 and 5 million metric tons (MT) by the year 2020 and the global supply will be at a deficit of 1 million MT.
Reliable productivity is therefore crucial for both the cacao industry and the livelihood of millions of cacao producers around the world.
Why we need your help?
In current practice, pre-harvest cocoa fruit predictions are provided based on a manual count of fruit number.
However, the variability in cocoa fruit number was demonstrated to be such that the best case sampling practice was inadequate for reliable estimation. Those previous results highlight the need for accurate automated fruit detection and counting methods.
How image annotations will be used ?
Annotated images will be used to train and test a Deep neural network [Instance segmentation model] allowing automatic fruit detection in the images. This will be further used to predict and geolocalise cocoa yield based on images.
Where are the samples coming from?
The images were obtained with the main camera of a mobile device (Samsung Galaxy S10 SM-G973F).
The lighting conditions, and occlusion of the fruits were not controlled and images were taken without flash.
The images were acquired in cocoa fields located in Ivory Coast (July of 2019) and in Cameroun (October 2019).
I want to know more!
More information can be found at CIRAD - UMR SYSTEM