Systems and methods for detecting waste receptacles using convolutional neural networks
Systems and methods for detecting a waste receptacle, the system including a camera for capturing an image, a convolutional neural network, and processor. The convolutional neural network can be trained for identifying target waste receptacles. The processor can be mounted on the waste-collection vehicle and in communication with the camera and the convolutional neural network configured for using the convolutional neural network. The processor can be configured for using the convolutional neural network to generate an object candidate based on the image; using the convolutional neural network to determine whether the object candidate corresponds to a target waste receptacle; and selecting an action based on whether the object candidate is acceptable.
1. A system for detecting a waste receptacle, comprising:
a camera configured to capture an image;
a convolutional neural network trained for identifying waste receptacles, the convolutional neural network comprising a plurality of depthwise separable convolution filters and a MobileNet architecture; and
one or more processors in communication with a waste-collection vehicle, the camera, and the convolutional neural network, the one or more processors configured to:
determine, based on use of the convolutional neural network, whether the image includes the waste receptacle; and
determine, based on whether the image includes the waste receptacle, an action.
2. The system of claim 1 , wherein determining, based on use of the convolutional neural network, whether the image includes the waste receptacle comprises predicting an object classification and a bounding box based on the image.
3. The system of claim 2 , wherein determining, based on use of the convolutional neural network, whether the image includes the waste receptacle further includes:
predicting, using the bounding box, a class confidence score; and
if the class confidence score is greater than a pre-defined confidence threshold of acceptability, determining that the image includes the waste receptacle;
otherwise determining that the waste receptacle is absent from the image.
4. The system of claim 2 , wherein the object classification comprises at least one of garbage, recycling, compost, and background.
5. The system of claim 2 , wherein the bounding box comprises pixel coordinates, a bounding box width, and a bounding box height.
6. The system of claim 1 , wherein the convolutional neural network comprises a meta-architecture for object classification and bounding box regression.
7. The system of claim 6 , wherein the meta-architecture comprises single shot detection.
8. The system of claim 7 , wherein the meta-architecture comprises four additional convolution layers.
9. The system of claim 1 , further comprising:
an arm for grasping the waste receptacle, the arm being mountable on the waste-collection vehicle; and
an arm-actuation module connected to the arm, the arm-actuation module in communication with the one or more processors;
wherein the one or more processors are further configured to:
select the action of picking up the waste receptacle if the image includes the waste receptacle;
select the action of rejecting an object included in the image; and
if the action of picking up the waste receptacle is selected, calculating a location of the waste receptacle; and
wherein the arm-actuation module is configured to automatically move the arm in response to the location of the waste receptacle.
10. The system of claim 9 , wherein the arm-actuation module is configured so that automatically moving the arm comprises grasping the waste receptacle.
11. The system of claim 10 , wherein moving the arm further comprises lifting the waste receptacle and dumping contents of the waste receptacle into the waste-collection vehicle.
12. A method for detecting a waste receptacle, comprising:
capturing an image with a camera;
determining, by one or more processors using a convolutional neural network, whether the image includes the waste receptacle, wherein the convolutional neural network comprises a plurality of depthwise separable convolution filters and a MobileNet architecture; and
determining, by the one or more processors, an action based on whether the image includes the waste receptacle.
13. The method of claim 12 , wherein determining, by the one or more processors using the convolutional neural network, whether the image includes the waste receptacle comprises predicting an object classification and a bounding box based on the image.
14. The method of claim 13 , wherein determining, by the one or more processors using the convolutional neural network, whether the image includes the waste receptacle further includes:
predicting, using the bounding box, a class confidence score; and
if the class confidence score is greater than a pre-defined confidence threshold of acceptability, determining that the image includes the waste receptacle;
otherwise determining that the waste receptacle is absent from the image.
15. The method of claim 13 , wherein the object classification comprises at least one of garbage, recycling, compost, or background.
16. The method of claim 13 , wherein the bounding box comprises pixel coordinates, a bounding box width, and a bounding box height.
17. The method of claim 12 , wherein the convolutional neural network comprises a meta-architecture for object classification and bounding box regression.
18. The method of claim 17 , wherein the meta-architecture comprises single shot detection.
19. The method of claim 18 , wherein the meta-architecture comprises four additional convolution layers.
20. The method of claim 12 , wherein determining the action based on whether the image includes the waste receptacle comprises:
selecting the action of picking up the waste receptacle if the image includes the waste receptacle; and
selecting the action of rejecting an object included in the image; and
if the action of picking up the waste receptacle is performed, the method further comprises:
calculating a location of the waste receptacle; and
moving an arm mounted on a waste-collection vehicle in response to the location of the waste receptacle.
21. The method of claim 20 , wherein moving the arm comprises grasping the waste receptacle.
22. The method of claim 21 , wherein moving the arm further comprises lifting the waste receptacle and dumping contents of the waste receptacle into the waste-collection vehicle.
23. A system for detecting a waste receptacle, comprising:
a camera configured to capture an image;
a convolutional neural network trained for identifying waste receptacles, the convolutional neural network comprising a plurality of depthwise separable convolution filters and a meta-architecture for object classification and bounding box regression; and
one or more processors in communication with a waste-collection vehicle, the camera, and the convolutional neural network, the one or more processors configured to:
determine, based on use of the convolutional neural network, whether the image includes the waste receptacle; and
determine, based on whether the image includes the waste receptacle, an action.