Deepcell

🐛 Bug Fixes

<details>
<summary>Add additional metrics to report in classify_divisions msschwartz21 (92)</summary>

Addresses 91 and bumps version to 0.5.6 for the next patch release after this PR is complete. I tested the new functionality in the model-registry and those updates can be seen in this branch: https://github.com/vanvalenlab/model-registry/compare/mrgn/tracking-evaluation. I added rounding after running this test, but decimals will now be truncated to 2 digits.
</details>


🧰 Maintenance

<details>
<summary>Bump copyright to 2022 and improve error message msschwartz21 (90)</summary>


</details>

🐛 Bug Fixes

<details>
<summary>Update `utils.trk_stats` to return a dictionary of stats as output msschwartz21 (89)</summary>

The `trk_stats` function originally printed stats, but did not return them to the user. This PR updates the function to return a dictionary of statistics and makes it possible to input `X`, `y` and `lineages` instead of loading the data from `filename`.
</details>

🐛 Bug Fixes

<details>
<summary>Fix bug in `is_valid_lineage`; daughters can be \< parent. willgraf (88)</summary>

No need to check if the daughters are in `all_cells`, as we check that for each lineage. Just check that the daughter is in the lineage.

Don't return False as a shortcut, just warn and continue to the next label. This will enable all warnings to be shown.
</details>

🐛 Bug Fixes

<details>
<summary>Remove `pathlib` from `setup.py`; fix build issues zverozabr (86)</summary>

Fixes 85

</details>


🧰 Maintenance

<details>
<summary>Bump version to 0.5.3. willgraf (87)</summary>


</details>

🐛 Bug Fixes

<details>
<summary>Speed up `trks_stats` by a factor of 5 willgraf (84)</summary>

`trks_stats` is used to calculate various statistics on a trks file. The old implementation ended up calling `np.unique` inside a double for loop, which caused the runtime to spike. These changes update the internal logic to greatly improve the runtime (namely, by separating the logic into two different for loops, updating the time complexity from O(batch * frames * cells) to O(batch*(frames+cells)).

Some example times on the current registry tracking data (all times include time to run `load_trks`):

- `test.trks` from 227s to 47s
- `val.trks` from 549s to 110s
- `train.trks` from 1968s to 386s
</details>

🐛 Bug Fixes

<details>
<summary>Use `np.any` instead of `np.sum` when finding all frames that contain a given object. willgraf (82)</summary>

`np.sum` counts all the frames the object is in, but we really only care that it exists in the given frame, not the number. `np.any` is about 3x faster (in my examples lineage), and should help speed up the slowest code in both `is_valid_lineage` and `relabel_sequential_lineage`.

Additionally, `is_valid_lineage` has some improved warning messages, and saves a single loop across the cells by iteratively removing the cells with the lineage from a set of all cells.
</details>


🧰 Maintenance

<details>
<summary>Bump the version to 0.5.1. willgraf (83)</summary>


</details>