Machine learning pill identification
Disclosed are various embodiments for automated pill identification using lighting devices and machine learning routines. A computing device may selectively control illumination of a pill provided at an imaging position by a pill dispensing system. The computing device may direct an imaging device to capture image data of the pill during illumination of the pill. Also, the computing device may generate a digital fingerprint of the pill and determine an identity of the pill based at least in part on a comparison of the digital fingerprint to a digital fingerprint library. A machine learning routine may be applied to improve future detection of the identity of the pill.
1. A system, comprising:
a pill dispensing system comprising at least one chamber configured to move a pill from an original position to an imaging position;
a first lighting device configured to illuminate the pill residing in imaging position from underneath the pill;
an imaging device positioned to capture image data of the pill residing in the imaging position;
a second lighting device configured to illuminate the pill while in the imaging position from above the pill;
at least one computing device being configured to:
selectively control at least one of the first lighting device to illuminate the pill from underneath;
direct the imaging device to capture first image data of the pill during illumination of the pill by the first lighting device;
selectively control at least one of the second lighting device to illuminate the pill from above;
direct the imaging device to capture second image data of the pill during illumination of the pill by the second lighting device;
generate a digital fingerprint of the pill, wherein the digital fingerprint comprises a red-green-blue histogram and a pill edge morphology generated based at least in part on the first image data and the second image data;
determine an identity of the pill based at least in part on a comparison of the digital fingerprint to a digital fingerprint library; and
apply a machine learning routine to improve future detection of the identity of the pill.
2. The system of claim 1 , wherein the first light emitting device comprises a substantially flat light emitting diode (LED) backlight; and
wherein the second lighting device comprises a plurality of light emitting diodes (LEDs).
3. The system of claim 1 , wherein the at least one computing device is further configured to generate a data object for the pill, wherein the data object comprises a plurality of variables and a plurality of values corresponding to the variables, wherein at least one of the plurality of variables comprises: a pill name, a pill area, a pill perimeter, a pill full area, a pill contour, and pill histogram data.
4. The system of claim 3 , wherein the pill histogram data comprises data corresponding to at least one red value, at least one green value, and at least one blue value.
5. A system, comprising:
at least one computing device;
program instructions stored in memory and executable by the at least one computing device that, when executed, direct the at least one computing device to:
selectively control illumination of a pill provided at an imaging position by a pill dispensing system;
direct an imaging device to capture image data of the pill during illumination of the pill from a position above the pill and illumination of the pill from a position beneath the pill;
generate a digital fingerprint of the pill comprising a histogram, the histogram generated based at least in part on the image data captured by the imaging device; and
determine an identity of the pill based at least in part on a comparison of the digital fingerprint to a digital fingerprint library.
6. The system of claim 5 , wherein the histogram is a red-green-blue histogram generated based at least in part on the image data captured by the imaging device.
7. The system of claim 6 , wherein the digital fingerprint further comprises a pill edge morphology generated based at least in part on the image data captured by the imaging device.
8. The system of claim 7 , further comprising the pill dispensing system, the pill dispensing system comprising at least one chamber, the imaging position comprising a position within the at least one chamber.
9. The system of claim 8 , further comprising a first lighting device configured to illuminate the pill residing in the at least one chamber from a position underneath the pill.
10. The system of claim 9 , further comprising an imaging device positioned to capture image data of the pill residing in the at least one chamber of the pill dispensing system, wherein the imaging device comprises a second lighting device positioned thereon for illuminating the at least one chamber from a position above the pill.
11. The system of claim 10 , wherein:
the image data comprises at least a first image and a second image;
the first image is acquired by the imaging device when the pill is illuminated by the first lighting device from a position underneath the pill; and
the second image is acquired by the imaging device when the pill is illuminated by the second lighting device from a position above the pill.
12. The system of claim 11 , wherein:
the pill edge morphology is generated based at least in part on the first image; and
the red-green-blue histogram is generated based at least in part on the second image.
13. The system of claim 12 , wherein:
the second imaging device comprises a plurality of light emitting diodes (LEDs); and
the at least one computing device is further directed to adjust at least one of: a red value, a blue value, and a green value associated with an illumination of at least one of the plurality of LEDs.
14. The system of claim 5 , wherein the at least one computing device is further directed to apply a machine learning routine to improve future detection of the identity of the pill.
15. A computer-implemented method, comprising:
selectively controlling, by at least one computing device, illumination of a pill provided at an imaging position by a pill dispensing system;
directing, by the at least one computing device, an imaging device to capture image data of the pill during illumination of the pill from a position above the pill and illumination of the pill from a position beneath the pill;
generating, by the at least one computing device, a digital fingerprint of the pill comprising histogram data, the histogram data generated based at least in part on the image data captured by the imaging device;
determining, by the at least one computing device, an identity of the pill based at least in part on a comparison of the digital fingerprint to a digital fingerprint library; and
applying, by the at least one computing device, a machine learning routine to improve future detection of the identity of the pill.
16. The computer-implemented method of claim 15 , wherein the histogram data comprises data associated with a red-green-blue histogram generated based at least in part on the image data captured by the imaging device.
17. The computer-implemented method of claim 15 , wherein the digital fingerprint further comprises a pill edge morphology generated based at least in part on the image data captured by the imaging device.
18. The computer-implemented method of claim 15 , further comprising directing, by the at least one computing device, a first lighting device to illuminate the pill from a position underneath the pill.
19. The computer-implemented method of claim 18 , further comprising directing, by the at least one computing device, a second lighting device positioned near an imaging device to illuminate the pill from a position above the pill.
20. The computer-implemented method of claim 19 , wherein:
the image data comprises at least a first image and a second image;
the first image is acquired by the imaging device when the pill is illuminated by the first lighting device from the position underneath the pill;
the second image is acquired by the imaging device when the pill is illuminated by the second lighting device from the position above the pill;
the pill edge morphology is generated based at least in part on the first image; and
the red-green-blue histogram is generated based at least in part on the second image.