Abstract: Accurately estimating crop yield is important to commercial grape production because it informs many vineyard and winery decisions. The current process of yield estimation is time consuming, is invasive to the crop and varies in its accuracy from 75-90\% depending on the experience of the viticulturist. This paper proposes a 4-channel input, 5-channel output, Multiple Task Learning (MTL) neural network approach. It uses images captured by inexpensive smart phones secured in a simple tripod arrangement. The MTL convolutional neural network approach develops models that achieved 85% accuracy from image data captured 6 days prior to harvest and 82% accuracy from images capture 16 days prior to harvest.