Seed germination detection method and apparatus

1. A method of determining seed germination on a soilless growth medium, comprising:

capturing a first image of one or more seeds, germinated seeds, or any combination of these exposed to germination conditions atop the soilless growth medium and converting the first image into a binary image;

identifying initial features corresponding to the one or more seeds, germinated seeds, or any combination of these in the binary image, each initial feature identified in the binary image by a number of interconnected white or black pixels above a threshold value; determining coordinates of a rectangle that encloses the interconnected white or black pixels above the threshold value; applying the coordinates of each said rectangle to the first image and cropping the initial features from the first image to form initial cropped feature images,

determining an average area of the initial features in the initial cropped feature images; dividing each initial cropped feature image by a scaled value of the average area of the initial features to form one or more final cropped feature images;

inputting each final cropped feature image from the first image into a deep learning model previously trained on images of germinated and non-germinated seeds;

determining whether each final cropped feature image from the first image is a germinated seed or a non-germinated seed based on an output of the deep learning model.

2. The method of claim 1 , further comprising generating a germination profile comprising a number of germinated seeds, non-germinated seeds, or any combination of these atop the soilless growth medium.

3. The method of claim 1 , wherein said deep learning model comprises a convolution neural network.

4. The method of claim 1 , wherein said deep learning model comprises convolution layers, dense layers, activation layers, or any combination of these.

5. The method of claim 1 , wherein the soilless growth medium is a textured cloth, fabric, or textile, said cloth, fabric, or textile has a texture on a scale similar to the size of the seeds.

6. The method of claim 1 , wherein the textured cloth comprises a napped surface or an outwardly directed napped surface.

7. The method of claim 1 , wherein the soilless growth medium comprises a textured fabric that is a loose woven material or a non-woven porous material.

8. The method of claim 1 , said deep learning model previously trained on images of germinated and non-germinated seeds atop the soilless growth medium.

9. The method of claim 1 , further comprising converting an HSV model of the first image into the binary image.

10. The method of claim 1 , further comprising determining whether to position the soilless growth medium in a growth chamber based on the germination profile.

11. The method of claim 1 , wherein the threshold number of interconnected white or black pixels is in a range from 100 pixels to 150 pixels.

12. A system for monitoring seed germination on a soilless growth medium, comprising:

a soilless growth medium comprising one or more seeds, germinating seeds, or any combination of these atop the soilless growth medium;

an image capturing device positioned to measure a germination status of the one or more seeds on the soilless growth medium exposed to germination conditions; and

a processor coupled to the image capturing device, the processor is further operable to:

capture a first image of one or more seeds, germinated seeds, or any combination of these exposed to germination conditions atop the soilless growth medium and convert the first image into a binary image;

identify initial features corresponding to the one or more seeds, germinated seeds, or any combination of these in the binary image, and crop the corresponding initial features from the first image to form initial cropped feature images;

determine an average area of the initial features in the initial cropped feature images; divide each initial cropped feature image by a scaled value of the average area of the initial features to form one or more final cropped feature images;

input each final cropped feature image into a deep learning model trained on images of germinated and non-germinated seeds; and,

calculate a number of germinated seeds, non-germinated seeds, or any combination of these atop the soilless growth medium based on an output of the deep learning model.

13. The system of claim 12 , wherein the processor is further operable to create a germination profile and determine whether to place the textured soilless growth medium in a grow chamber based on the germination profile.

14. The system of claim 12 , wherein said deep learning model comprises convolution layers and dense layers.

15. The system of claim 12 , wherein the soilless growth medium is a textured cloth, fabric, or textile, said cloth, fabric, or textile has a texture on a scale similar to the size of the seeds.

16. The system of claim 11 , wherein the soilless growth medium is a fabric that has a loose woven or is a non-woven porous substrate.

17. The system of claim 12 , wherein the soilless growth medium comprises a layer of a paper and layer of a loose woven or a non-woven porous substrate atop the layer of paper.

18. The system of claim 12 , wherein the soilless growth medium further includes stem fragments.

19. The system of claim 12 , wherein the soilless growth medium comprises an outwardly directed nap on both the top and bottom surfaces and stem fragments.

20. The system of claim 12 , wherein the soilless growth medium is positioned on a frame to form a flat, and wherein said flat further comprises a light barrier layer atop the soilless growth medium and seeds exposed to germination conditions are positioned in openings in the light barrier layer.

21. A computer program product comprising a set of computer instructions stored a non-transitory computer readable medium, the computer instructions comprising instructions executable by a processor to:

receive a first image of one or more seeds, germinated seeds, or any combination of these exposed to germination conditions atop a soilless growth medium and convert the first image into a black and white binary image;

capture a first image of one or more seeds, germinated seeds, or any combination of these exposed to germination conditions atop the soilless growth medium and convert the first image into a binary image;

identify initial features corresponding to the one or more seeds, germinated seeds, or any combination of these in the binary image, and crop the corresponding initial features from the first image to form initial cropped feature images;

determine an average area of the initial features in the initial cropped feature images; divide each initial cropped feature image by a scaled value of the average area of the initial features to form one or more final cropped feature images;

input each final cropped feature image into a deep learning model trained on images of germinated and non-germinated seeds; and,

calculate a number of germinated seeds, non-germinated seeds, or any combination of these atop the soilless growth medium based on an output of the deep learning model.

22. The computer program product of claim 21 , wherein said deep learning model further comprises convolution layers and dense layers.

23. The computer program product of claim 21 , wherein the computer instructions are further executable by the processor to determine whether to position the soilless growth medium in a growth chamber based on the germination profile.

24. A method of determining seed germination on a soilless growth medium, comprising:

capturing a first image of one or more seeds, germinated seeds, or any combination of these exposed to germination conditions atop the soilless growth medium and converting the first image into a binary image;

identifying initial features corresponding to the one or more seeds, germinated seeds, or any combination of these in the binary image, and cropping the corresponding initial features from the first image to form initial cropped feature images;

determining an average area of the initial features in the initial cropped feature images; dividing each initial cropped feature image by a scaled value of the average area of the initial features to form one or more final cropped feature images;

inputting each final cropped feature image into a deep learning model trained on images of germinated and non-germinated seeds; and,

calculating a number of germinated seeds, non-germinated seeds, or any combination of these atop the soilless growth medium based on an output of the deep learning model.

25. The method of claim 24 , wherein the initial features in the binary image are identified based on a number of interconnected white or black pixels above a threshold value.

26. The method of claims 24 , further comprising determining coordinates of a rectangle that encloses the interconnected white or black pixels above the threshold value; applying the coordinates of each rectangle to the first image and cropping initial features from the first image based on the coordinates of each said rectangle.

27. The method of claim 24 further comprising creating a germination profile based on the output from the deep learning model.

28. The method of claim 24 , wherein the identifying the initial features is based on threshold number of interconnected white or black pixels is in a range from 100 pixels to 150 pixels.

29. The method of claim 24 , wherein the deep learning model previously trained on images of germinated and non-germinated seeds atop a comparable soilless growth medium.