Sea lice mitigation based on historical observations

1. A computer-implemented method comprising:

obtaining, over a period of time, multiple observations of a first fish within a population of reference fish;

generating, from the multiple observations, a record for the first fish indicating an extent of sea lice infestation for the first fish at each observation of the multiple observations and whether the first fish was healthy when harvested;

training, based at least in part on the record for the first fish, a model that determines, given one or more input records for a given fish, whether the given fish is likely to be healthy when harvested;

obtaining an image of a sample fish that is not in the population of reference fish;

identifying a record that indicates an extent of sea lice infestation previously observed on the sample fish;

determining, based at least on inputting the extent of sea lice infestation previously observed on the sample fish to the model, whether the sample fish is likely to be healthy when harvested; and

selectively initiating sea lice mitigation based on determining whether the sample fish is likely to be healthy when harvested.

2. The method of claim 1 , wherein determining whether the sample fish is likely to be healthy when harvested comprises:

determining a current extent of sea lice infestation for the sample fish based on the image;

providing, to the model, a representation of the current extent of sea lice infestation for the sample fish; and

obtaining, from the model in response to providing the representation, a predicted future health indicator for the sample fish.

3. The method of claim 2 , wherein determining the current extent of sea lice infestation for the sample fish based on the image comprises:

determining a location of each sea lice on the sample fish,

wherein providing, to the model, the representation of the current extent of sea lice infestation for the sample fish comprises providing an indication of locations of each sea lice to the model.

4. The method of claim 2 , wherein identifying the record that indicates the extent of sea lice infestation previously observed on the sample fish comprises:

extracting visual features of the sample fish;

identifying the record that indicates the extent of sea lice infestation previously observed on the sample fish based on the visual features; and

providing, to the model, both a representation of the extent of sea lice infestation previously observed on the sample fish and the representation of the current extent of sea lice infestation for the sample fish based on the image.

5. The method of claim 1 , wherein selectively initiating the sea lice mitigation comprises:

providing, to a sea lice treatment device, an instruction to treat the sample fish for sea lice.

6. The method of claim 1 , wherein selectively initiating the sea lice mitigation comprises:

determining that a predicted future health indicator satisfies a mitigation criteria; and

in response to determining that the predicted future health indicator satisfies the mitigation criteria, initiating the sea lice mitigation.

7. The method of claim 1 , wherein selectively initiating the sea lice mitigation comprises:

determining that a predicted future health indicator does not satisfy a mitigation criteria; and

in response to determining that the predicted future health indicator does not satisfy the mitigation criteria, not initiating the sea lice mitigation.

8. The method of claim 1 , wherein the record for the first fish indicates one or more of age, weight, size, a feature vector, or various health metrics of the first fish at two or more different times across the period of time, and

determining whether the sample fish is likely to be healthy when harvested based on one or more of an age, weight, size, feature vector, or various health metrics of the sample fish at two or more other different times across another period of time.

9. The method of claim 1 , wherein the record for the first fish indicates conditions of environments in which the multiple observations were made, and

determining whether the sample fish is likely to be healthy when harvested based on a condition of an environment in which the image of the sample fish was obtained.

10. The method of claim 9 , wherein the conditions of environments in which the multiple observations were made includes one or more of location, depth, or temperature.

11. The method of claim 1 , wherein the record for the first fish indicates whether sea lice mitigation was performed for the first fish.

12. The method of claim 1 , wherein generating the record for the first fish comprises:

determining a number of sea lice on the first fish based on an observation of the multiple observations; and

storing, for each of the multiple observations, a corresponding determined number of sea lice on the first fish in the record.

13. The method of claim 1 , wherein obtaining the image of the sample fish that is not in the population of reference fish comprises:

obtaining a set of images of the sample fish that is not in the population of reference fish, where the set of images includes the image of the sample fish,

wherein determining, based on the image of the sample fish and with the model, whether the sample fish is likely to be healthy when harvested is based on the set of images of the sample fish.

14. A system comprising:

one or more computers and one or more storage devices storing instructions that are operable, when executed by the one or more computers, to cause the one or more computers to perform operations comprising:

obtaining, over a period of time, multiple observations of a first fish within a population of reference fish;

generating, from the multiple observations, a record for the first fish indicating an extent of sea lice infestation for the first fish at each observation of the multiple observations and whether the first fish was healthy when harvested;

training, based at least in part on the record for the first fish, a model that determines, given one or more input records for a given fish, whether the given fish is likely to be healthy when harvested;

obtaining an image of a sample fish that is not in the population of reference fish;

identifying a record that indicates an extent of sea lice infestation previously observed on the sample fish;

determining, based at least on inputting the extent of sea lice infestation previously observed on the sample fish to the model, whether the sample fish is likely to be healthy when harvested; and

selectively initiating sea lice mitigation based on determining whether the sample fish is likely to be healthy when harvested.

15. The system of claim 14 , wherein determining whether the sample fish is likely to be healthy when harvested comprises:

determining a current extent of sea lice infestation for the sample fish based on the image;

providing, to the model, a representation of the current extent of sea lice infestation for the sample fish; and

obtaining, from the model in response to providing the representation, a predicted future health indicator for the sample fish.

16. The system of claim 15 , wherein determining the current extent of sea lice infestation for the sample fish based on the image comprises:

determining a location of each sea lice on the sample fish,

wherein providing, to the model, the representation of the current extent of sea lice infestation for the sample fish comprises providing an indication of locations of each sea lice to the model.

17. The system of claim 15 , wherein identifying the record that indicates the extent of sea lice infestation previously observed on the sample fish comprises:

extracting visual features of the sample fish;

identifying the record that indicates the extent of sea lice infestation previously observed on the sample fish based on the visual features; and

providing, to the model, both a representation of the extent of sea lice infestation previously observed on the sample fish and the representation of the current extent of sea lice infestation for the sample fish based on the image.

18. The system of claim 14 , wherein selectively initiating the sea lice mitigation comprises:

providing, to a sea lice treatment device, an instruction to treat the sample fish for sea lice.

19. The system of claim 14 , wherein selectively initiating the sea lice mitigation comprises:

determining that a predicted future health indicator satisfies a mitigation criteria; and

in response to determining that the predicted future health indicator satisfies the mitigation criteria, initiating the sea lice mitigation.

20. A computer-readable storage device encoded with a computer program, the program comprising instructions that when executed by one or more computers cause the one or more computers to perform operations comprising:

obtaining, over a period of time, multiple observations of a first fish within a population of reference fish;

generating, from the multiple observations, a record for the first fish indicating an extent of sea lice infestation for the first fish at each observation of the multiple observations and whether the first fish was healthy when harvested;

training, based at least in part on the record for the first fish, a model that determines, given one or more input records for a given fish, whether the given fish is likely to be healthy when harvested;

obtaining an image of a sample fish that is not in the population of reference fish;

identifying a record that indicates an extent of sea lice infestation previously observed on the sample fish;

determining, based at least on inputting the extent of sea lice infestation previously observed on the sample fish to the model, whether the sample fish is likely to be healthy when harvested; and

selectively initiating sea lice mitigation based on determining whether the sample fish is likely to be healthy when harvested.