Inferring moisture from color
Techniques are described herein for using artificial intelligence to predict crop yields based on observational crop data. A method includes: obtaining a first digital image of at least one plant; segmenting the first digital image of the at least one plant to identify at least one seedpod in the first digital image; for each of the at least one seedpod in the first digital image: determining a color of the seedpod; determining a number of seeds in the seedpod; inferring, using one or more machine learning models, a moisture content of the seedpod based on the color of the seedpod; and estimating, based on the moisture content of the seedpod and the number of seeds in the seedpod, a weight of the seedpod; and predicting a crop yield based on the moisture content and the weight of each of the at least one seedpod.
1. A method implemented by one or more processors, the method comprising:
obtaining a first digital image of at least one plant;
segmenting the first digital image of the at least one plant to identify at least one seedpod in the first digital image;
for each of the at least one seedpod in the first digital image:
determining a color of the seedpod;
determining a number of seeds in the seedpod;
inferring, using one or more machine learning models, a moisture content of the seedpod based on the color of the seedpod, wherein:
at least one of the one or more machine learning models is a convolutional neural network model,
each of the one or more machine learning models is trained using a plurality of digital images of a plurality of plants, and
at least one of the plurality of digital images is labeled based on a ground truth seedpod moisture content; and
estimating, based on the moisture content of the seedpod and the number of seeds in the seedpod, a weight of the seedpod; and
predicting a crop yield based on the moisture content and the weight of each of the at least one seedpod.
2. The method according to claim 1 , further comprising determining a size of each of the at least one seedpod in the first digital image,
wherein, for each of the at least one seedpod in the first digital image, the estimating the weight of the seedpod is further based on the size of the seedpod.
3. The method according to claim 1 , further comprising determining a number of seedpods on each of the at least one plant in the first digital image,
wherein the predicting the crop yield is further based on the number of seedpods on each of the at least one plant in the first digital image.
4. The method according to claim 1 , wherein:
the weight that is estimated based on the moisture content is a wet weight; and
the predicting the crop yield comprises predicting a dry weight based on the wet weight of each of the at least one seedpod and the moisture content of each of the at least one seedpod.
5. The method according to claim 1 , wherein:
the first digital image is one of a plurality of digital images of the at least one plant obtained using a multi-camera array, and
the segmenting, the determining the color, the determining the number of seeds, the inferring, and the estimating are performed using the plurality of digital images.
6. The method according to claim 5 , wherein the plurality of digital images comprises digital images obtained at a plurality of positions along a length of a row of plants.
7. A system comprising:
a processor, a computer-readable memory, one or more computer-readable storage media, and program instructions collectively stored on the one or more computer-readable storage media, the program instructions executable to:
obtain a first digital image of at least one plant;
segment the first digital image of the at least one plant to identify at least one seedpod in the first digital image;
for each of the at least one seedpod in the first digital image:
determine a color of the seedpod;
determine a number of seeds in the seedpod; and
infer, using one or more machine learning models, one or both of a moisture content of the seedpod and a weight of the seedpod, based on the color of the seedpod and the number of seeds in the seedpod, wherein:
at least one of the one or more machine learning models is a convolutional neural network model,
each of the one or more machine learning models is trained using a plurality of digital images of a plurality of plants, and
at least one of the plurality of digital images is labeled based on a ground truth seedpod moisture content; and
predict a crop yield based on the moisture content and the weight of each of the at least one seedpod.
8. The system according to claim 7 , wherein:
the program instructions are further executable to determine a size of each of the at least one seedpod in the first digital image; and
for each of the at least one seedpod in the first digital image, the weight of the seedpod is determined based on the size of the seedpod.
9. The system according to claim 7 , wherein:
the program instructions are further executable to determine a number of seedpods on each of the at least one plant in the first digital image; and
the predicting the crop yield is further based on the number of seedpods on each of the at least one plant in the first digital image.
10. The system according to claim 7 , wherein:
the weight is a wet weight; and
the predicting the crop yield comprises predicting a dry weight based on the wet weight of each of the at least one seedpod and the moisture content of each of the at least one seedpod.
11. The system according to claim 7 , wherein:
the first digital image is one of a plurality of digital images of the at least one plant obtained using a multi-camera array, and
the segmenting, the determining the color, the determining the number of seeds, and the inferring are performed using the plurality of digital images.
12. The system according to claim 11 , wherein the plurality of digital images comprises digital images obtained at a plurality of positions along a length of a row of plants.