Aggregate trait estimation for agricultural plots
Implementations are described herein for estimating aggregate trait values for agricultural plots. In various implementations, a plurality of images depicting crops in an agricultural plot may be processed using a feature extraction portion of aggregate trait estimation model to generate a corresponding plurality of embeddings. The plurality of embeddings may be combined to obtain a unified representation of the crops in the agricultural plot. Using a prediction portion of the aggregate trait estimation model, the unified representation of the crops may be processed to estimate, as a direct output of the aggregate trait estimation model, an aggregate trait value of the crops in the agricultural plot. Data indicative of the estimated aggregate trait value of the crops in the agricultural plot may be output at a computing device.
1. A method implemented using one or more processors and comprising:
processing, using a feature extraction portion of a first aggregate trait estimation model, a first plurality of images depicting crops in an agricultural plot to generate a corresponding first plurality of embeddings;
processing, using a feature extraction portion of a second aggregate trait estimation model, a second plurality of images to generate a corresponding second plurality of embeddings;
combining the first plurality of embeddings to obtain a first unified representation of the crops in the agricultural plot;
combining the second plurality of embeddings to obtain a second unified representation of the crops in the agricultural plot;
processing, using prediction portions of the first and second aggregate trait estimation models, the first and second unified representations of the crops to estimate an aggregate trait value of the crops in the agricultural plot; and
causing data indicative of the estimated aggregate trait value of the crops in the agricultural plot to be output at a computing device.
2. The method of claim 1 , further including capturing the first plurality of images using a first vision sensor carried across the agricultural plot by an agricultural vehicle.
3. The method of claim 2 , further including capturing the second plurality of images using a second vision sensor carried across the agricultural plot by the agricultural vehicle, wherein the first vision sensor captures the crops from a first angle and the second vision sensor captures the crops from a second angle that is different than the first angle.
4. The method of claim 3 , wherein the first angle includes a side view of the crops and the second angle comprises a top-down view of the crops.
5. The method of claim 3 , wherein the aggregate trait value is a final aggregate trait value, and the method further includes:
processing, using the prediction portion of the first aggregate trait estimation model, the first unified representation of the crops to estimate a first aggregate trait value of the crops in the agricultural plot; and
processing, using the prediction portion of the second aggregate trait estimation model, the second unified representation of the crops to estimate a second aggregate trait value of the crops in the agricultural plot.
6. The method of claim 5 , further including calculating the final aggregate trait value of the crops in the agricultural plot based on the first and second aggregate trait values.
7. The method of claim 1 , wherein the combining includes concatenating or averaging the first plurality of embeddings.
8. The method of claim 1 , wherein the estimated aggregate trait value includes a probability within a normalized range that corresponds to non-normalized, ground truth aggregate trait values observed in a plurality of fields.
9. A system comprising one or more processors and memory storing instructions that, in response to execution of the instructions, cause the one or more processors to:
process, using a feature extraction portion of a first aggregate trait estimation model, a first plurality of images depicting crops in an agricultural plot to generate a corresponding first plurality of embeddings;
process, using a feature extraction portion of a second aggregate trait estimation model, a second plurality of images to generate a corresponding second plurality of embeddings;
combine the first plurality of embeddings to obtain a first unified representation of the crops in the agricultural plot;
combine the second plurality of embeddings to obtain a second unified representation of the crops in the agricultural plot;
process, using prediction portions of the first and second aggregate trait estimation models, the first and second unified representations of the crops to estimate an aggregate trait value of the crops in the agricultural plot; and
cause data indicative of the estimated aggregate trait value of the crops in the agricultural plot to be output at a computing device.
10. The system of claim 9 , further including instructions to capture the first plurality of images using a first vision sensor carried across the agricultural plot by an agricultural vehicle.
11. The system of claim 10 , further including instructions to capture the second plurality of images using a second vision sensor carried across the agricultural plot by the agricultural vehicle, wherein the first vision sensor captures the crops from a first angle and the second vision sensor captures the crops from a second angle that is different than the first angle.
12. The system of claim 11 , wherein the first angle includes a side view of the crops and the second angle comprises a top-down view of the crops.
13. The system of claim 11 , wherein the aggregate trait value is a final aggregate trait value, and the memory includes instructions to:
process, using the prediction portion of the first aggregate trait estimation model, the first unified representation of the crops to estimate a first aggregate trait value of the crops in the agricultural plot; and
process, using the prediction portion of the second aggregate trait estimation model, the second unified representation of the crops to estimate a second aggregate trait value of the crops in the agricultural plot.
14. The system of claim 13 , further including instructions to calculate the final aggregate trait value of the crops in the agricultural plot based on the first and second aggregate trait values.
15. The system of claim 9 , wherein the first plurality of embeddings are combined by concatenation or averaging.
16. The system of claim 9 , wherein the estimated aggregate trait value includes a probability within a normalized range that corresponds to non-normalized, ground truth aggregate trait values observed in a plurality of fields.
17. A non-transitory computer-readable medium comprising instructions that, in response to execution of the instructions by a processor, cause the processor to:
process, using a feature extraction portion a first aggregate trait estimation model, a first plurality of images depicting crops in an agricultural plot to generate a corresponding first plurality of embeddings;
process, using a feature extraction portion of a second aggregate trait estimation model, a second plurality of images to generate a corresponding second plurality of embeddings;
combine the first plurality of embeddings to obtain a first unified representation of the crops in the agricultural plot;
combine the second plurality of embeddings to obtain a second unified representation of the crops in the agricultural plot;
process, using prediction portions of the first and second aggregate trait estimation models, the first and second unified representations of the crops to estimate an aggregate trait value of the crops in the agricultural plot; and
cause data indicative of the estimated aggregate trait value of the crops in the agricultural plot to be output at a computing device.
18. The non-transitory computer-readable medium of claim 17 , further including instructions to capture the first plurality of images using a first vision sensor carried across the agricultural plot by an agricultural vehicle.
19. The non-transitory computer-readable medium of claim 18 , further including instructions to capture the second plurality of images using a second vision sensor carried across the agricultural plot by the agricultural vehicle, wherein the first vision sensor captures the crops from a first angle and the second vision sensor captures the crops from a second angle that is different than the first angle.
20. The non-transitory computer-readable medium of claim 19 , wherein the first angle includes a side view of the crops and the second angle comprises a top-down view of the crops.