Fish biomass, shape, and size determination
Methods, systems, and apparatuses, including computer programs encoded on a computer-readable storage medium for estimating the shape, size, and mass of fish are described. A pair of stereo cameras may be utilized to obtain right and left images of fish in a defined area. The right and left images may be processed, enhanced, and combined. Object detection may be used to detect and track a fish in images. A pose estimator may be used to determine key points and features of the detected fish. Based on the key points, a three-dimensional (3-D) model of the fish is generated that provides an estimate of the size and shape of the fish. A regression model or neural network model can be applied to the 3-D model to determine a likely weight of the fish.
1. A method for positioning one or more cameras within a fish pen, the method comprising:
receiving, at one or more computing devices, information identifying a type of fish housed within the fish pen;
identifying one or more swimming patterns associated with the type of fish; and
positioning the one or more cameras within the fish pen based on the one or more swimming patterns associated with the type of fish.
2. The method of claim 1 , wherein the one or more swimming patterns associated with the type of fish include information identifying a swimming direction relative to a direction of water current associated with the type of fish, and the one or more cameras are positioned such that a longitudinal axis of a fish of the type of fish traveling along the swimming direction is substantially perpendicular to a view direction of the one or more cameras.
3. The method of claim 1 , wherein the one or more swimming patterns associated with the type of fish include information identifying a depth at which the type of fish prefers to swim, and the one or more cameras are positioned at the corresponding depth.
4. The method of claim 1 , wherein the one or more swimming patterns associated with the type of fish include information identifying a temperature range at which the type of fish prefers to swim, and positioning the one or more cameras comprises:
identifying a region within the fish pen where the water temperature is within the identified temperature range; and
positioning the one or more cameras to image fish of the type of fish swimming through the identified region.
5. The method of claim 1 , wherein the one or more swimming patterns associated with the type of fish include information identifying an illumination range at which the type of fish prefers to swim, and positioning the one or more cameras comprises:
identifying a region within the fish pen where the amount of light is within the identified illumination range; and
positioning the one or more cameras to image fish of the type of fish swimming through the identified region.
6. The method of claim 1 , further comprising:
capturing at least one image using the one or more cameras positioned within the fish pen in accordance with the one or more swimming patterns associated with the type of fish;
identifying a fish of the type of fish within the at least one image;
determining that an orientation of the fish relative to the camera does not satisfy a threshold condition; and
adjusting corresponding locations of the one or more cameras responsive to determining that the orientation of the fish relative to the camera does not satisfy the threshold condition.
7. The method of claim 1 , wherein the information identifying the type of fish is received at the one or more computing devices in the form of a user-input.
8. A system comprising:
one or more cameras configured to capture underwater images in a fish pen;
one or more computing devices operably coupled to a control system configured to control the one or more cameras; and
one or more storage devices storing instructions which when executed by the one or more computing devices, cause the one or more computing devices to perform operations comprising:
receiving, at one or more computing devices, information identifying a type of fish housed within the fish pen;
identifying one or more swimming patterns associated with the type of fish; and
positioning the one or more cameras within the fish pen based on the one or more swimming patterns associated with the type of fish.
9. One or more non-transitory computer-readable storage media comprising instructions, which, when executed by one or more computing devices, cause the one or more computing devices to perform operations comprising:
receiving, at one or more computing devices, information identifying a type of fish housed within the fish pen;
identifying one or more swimming patterns associated with the type of fish; and
positioning the one or more cameras within the fish pen based on the one or more swimming patterns associated with the type of fish.
10. A method for uniquely identifying a fish within a fish pen, the method comprising:
obtaining, using one or more cameras, a first image of a region of the fish pen;
generating a bounding box around a fish that is visible in the first image; and
identifying, within the bounding box around the fish that is visible in the first image, one or more spot patterns on the fish; and
generating a unique identification of the fish based at least on the one or more spot patterns.
11. The method of claim 10 , further comprising:
computing, from the first image, one or more characteristics of the fish; and
storing, in a database as a part of time series data, information representing the one or more characteristics computed from the first image in association with the unique identification of the fish.
12. The method of claim 11 , further comprising:
obtaining at least a second image, wherein the second image is captured before or after the first image;
detecting, within the second image, the one or more spot patterns on the fish;
computing, from the second image, the one or more characteristics of the fish; and
storing, in the database as part of the time series data, information representing the one or more characteristics computed from the second image in association with the unique identification of the fish.
13. The method of claim 11 , wherein the one or more characteristics include at least one of: a weight, a size, or a length of the fish.
14. The method of claim 11 , wherein a region of the fish pen captured in the second image is different from the region of the fish pen captured in the first image.
15. The method of claim 10 , wherein the bounding box is generated via object detection by a trained recurrent convolutional neural network (RCNN), or semantic segmentation that segments the portion from a background.