Grain quality monitoring
A method and non-transitory computer-readable medium capture an image of bulk grain and apply a feature extractor to the image to determine a feature of the bulk grain in the image. For each of a plurality of different sampling locations in the image, based upon the feature of the bulk grain at the sampling location, a determination is made regarding a classification score for the presence of a classification of material at the sampling location. A quality of the bulk grain of the image is determined based upon an aggregation of the classification scores for the presence of the classification of material at the sampling locations.
1. A method comprising: capturing an image of bulk grain; applying a feature extractor to the image to determine a feature of the bulk grain in the image; for each of a plurality of different sampling locations within the image, determining, based upon the feature of the bulk grain at the sampling location, a classification score for presence of a classification of material at the sampling location; and outputting a signal indicating quality of the bulk grain of the image based upon an aggregate of the classification scores of the different sampling locations; for each of the plurality of different sampling locations in the image, determining, based upon the plurality of features of the bulk grain at the sampling location, a classification score for the presence of a second classification of material at the sampling location, wherein the signal indicating quality of the bulk grain of the image is based upon an aggregate of the classification scores for the presence of the second classification of material at each of the sampling locations, and adjusting operational settings of a harvester based upon the signal.
2. The method of claim 1 comprising applying a plurality of feature extractors to the image to determine a plurality of features of the bulk grain in the image, wherein the classification score for the presence of the classification of material at the sampling location is based upon the plurality of features of the bulk grain at the sampling location.
3. The method of claim 1 further comprising:
segmenting the image into segments defined by presumed boundaries of individual bulk grain elements;
determining a feature of each segment, wherein the classification score for the presence of the classification of material at each sampling location is additionally based upon the feature of each segment at each sampling location.
4. The method of claim 1 further comprising:
segmenting the image into segments defined by presumed boundaries of individual bulk grain elements;
applying a plurality of feature extractors to each of the segments to determine a plurality of features for each segment, wherein the classification score for the presence of the classification of material at each sampling location is additionally based upon the plurality of features of each segment at each sampling location.
5. The method of claim 1 further comprising:
for each of a plurality of different sampling locations in the image, determining, based upon the feature of the bulk grain at the sampling location, a classification score for the presence of a second classification of material at the sampling location; and
outputting a signal indicating quality of the bulk grain of the image based upon an aggregate of the classification scores for the presence of the second classification at each of the sampling locations.
6. The method of claim 1 further comprising outputting a signal indicating a confidence level in the signal indicating quality of the bulk grain of the image.
7. The method of claim 1 , wherein the classification of material is selected from a group of classifications of material consisting of: clean grain, broken grain, material other than grain-light, material other than grain-heavy, and unthreshed grain.
8. The method of claim 1 further comprising applying a calibration function to the aggregate of the classification scores to determine a percent by weight of material having the classification for the image, wherein the signal is based upon the determined percent by weight of the material having the classification for the image.
9. The method of claim 1 , wherein the signal is based upon a time or material flow weighted filter of a plurality of the aggregate of classification scores for the classification from multiple images of bulk grain.
10. A non-transitory computer-readable medium containing program logic to direct a processor to: receive signals corresponding to a captured image of bulk grain, the captured image comprising (A) multiple individual bulk grain elements that are partially overlapped by other bulk grain elements or that are partially overlapping other bulk grain elements and (B) non-grain elements selected from a group of non-grain elements consisting of: chaff, hulls, cobs, stems and pods; apply a feature extractor to the image to determine a feature of the bulk grain in the image; for each of a plurality of different sampling locations in the image, determine, based upon the feature of the bulk grain at the sampling location, a classification score for the presence of a classification of material at the sampling location; and output a signal indicating quality of the bulk grain of the image based upon an aggregate of the classification scores for the presence of the classification of material at each of the different sampling locations; for each of the plurality of different sampling locations in the image, determining, based upon the plurality of features of the bulk grain at the sampling location, a classification score for the presence of a second classification of material at the sampling location, wherein the signal indicating quality of the bulk grain of the image is based upon an aggregate of the classification scores for the presence of the second classification of material at each of the sampling locations, wherein the signal is based upon a time or material flow weighted filter of a plurality of the aggregate of classification scores for the classification from multiple images and adjusting operational settings of a harvester based upon the signal.
11. The non-transitory computer-readable medium of claim 10 , wherein the program logic is further configured to direct the processor to:
segment the image into segments defined by presumed boundaries of individual bulk grain elements;
apply a feature extractor to the segments to determine a feature of each segment, wherein the classification score for the presence of the classification of material at the sampling location is additionally based upon the feature of each segment at the sampling location.
12. The non-transitory computer-readable medium of claim 10 , wherein the program logic is further to direct the processor to:
segment the image into segments defined by presumed boundaries of individual bulk grain elements;
apply a plurality of feature extractors to each of the segments to determine a plurality of features for each segment, wherein the classification score for the presence of the classification of material at the sampling location is additionally based upon the plurality of features of each segment at the sampling location.
13. The non-transitory computer-readable medium of claim 10 , wherein the program logic is further to direct the processor to:
apply a plurality of feature extractors to the image to determine a plurality of features of the bulk grain in the image; and
for each of the plurality of different sample locations in the image, determine, based upon the plurality of features of the bulk grain at the sampling location, a classification score for presence of a second classification of material at the sampling location, wherein the signal indicating quality of the bulk grain of the image is based upon an aggregate of the classification scores for the presence of a second classification of material at each of the sampling locations.
14. The non-transitory computer-readable medium of claim 10 , wherein the program logic is further to direct the processor to apply a calibration function to the aggregate of the classification scores to determine a percent by weight of the classification of material for the image, wherein the signal is based upon the determined percent by weight of the classification of material for the image.
15. The non-transitory computer-readable medium of claim 10 , wherein the program logic is further to direct the processor to output a signal indicating a confidence level in the signal indicating quality of the bulk grain of the image.
16. The non-transitory computer-readable medium of claim 10 , wherein the program logic is further to direct the processor to output a signal indicating an uncertainty in the signal indicating quality of the bulk grain of the image.
17. The non-transitory computer-readable medium of claim 1 , wherein the harvester harvests grain from plants in a growing medium.
18. The method of claim 1 , wherein the image of bulk comprises (A) multiple individual bulk grain elements that are partially overlapped by other bulk grain elements or that are partially overlapping other bulk grain elements and (B) non-grain elements selected from a group of non-grain elements consisting of: chaff, hulls, cobs, stems and pods.
19. The method of claim 1 , wherein the automatic adjustment of the operational settings of the harvester based upon the signal is carried out as a harvester is traversing a field and as the bulk grain is being harvested by the harvester.
20. A method comprising: capturing an image of bulk grain; applying a feature extractor to the image to determine a feature of the bulk grain in the image; for each of a plurality of different sampling locations within the image, determining, based upon the feature of the bulk grain at the sampling location, a classification score for presence of a classification of material at the sampling location; outputting a signal indicating quality of the bulk grain of the image based upon an aggregate of the classification scores of the different sampling locations; and applying a calibration function to the aggregate of the classification scores to determine a percent by weight of material having the classification for the image, wherein the signal is based upon the determined percent by weight of the material having the classification for the image; for each of the plurality of different sampling locations in the image, determining, based upon the plurality of features of the bulk grain at the sampling location, a classification score for the presence of a second classification of material at the sampling location, wherein the signal indicating quality of the bulk grain of the image is based upon an aggregate of the classification scores for the presence of the second classification of material at each of the sampling locations, and adjusting operational settings of a harvester based upon the signal.