Cornell Orchards

Yield estimation is an important indication in determining a particular farm/vineyard's success. However, in large farms, the current method of estimating the yield is by counting some random samples of produce and extrapolating the value based on the field size. This method is generally expensive and has high uncertainty. Riesling grape cluster estimation project involves making an app estimating grape clusters using cell phone imaging. As a picture of a vine is taken, the program automatically detects the clusters in the picture and gives the count. It even gives a low-error estimation on the clusters which are hiding behind leaves, wires, or which are coiled. All of this estimation is based on the data which it depends on to make the decision. This means that the program is trained to detect the cluster through hundreds of cluster images which are previously identified as clusters. This method of estimation provides farmers an easy, cheap as well as a fast way of estimating the yield by just using their own cellphone and walking/driving across the fields. 

I was involved in training the neural network to identify Riesling grape clusters and shoot tips on field data collected in Lansing, NY. Shoot tips also give a good estimation of the cluster number which will be growing from it in the future. labelIMG was used in order to identify clusters and shoots in the images of vines. 

Lansing's vineyard consisted of around 35,000 clusters and 12000 shoots of 500 grapevines. The images of all of this were taken in different time intervals of growth to maximize the proper detection of the clusters. The estimation of cluster number by the program had to be verified by individual counting of the clusters in the vineyard. 

Agriculture techniques from engineering approach: 

Working in the orchards, I came across several techniques adapted to grow trees. Most of them followed simple energy conservation. Energy balance was performed between the energy supplied from fertilizers, sun, and other forms with energy that the plant uses for its growth and production of fruits. As a farmer, the main goal is to produce maximum fruits. Therefore, there are several techniques developed which allow the plant to spend minimum energy in trunk and branches, and maximum in the fruits for higher yield. Upright Fruiting Offshoots (UFO) system is one of them which is widely used in the cherry trees. However, for sustainable fruit production, trying to supply maximum energy to the fruits does not always work. It turns out if most of the plants' energy is used, the yield would decrease in the following years. Therefore, it is important to have a good amount of research and analysis to figure out what might be the most efficient and sustainable way for fruit production.  