|
2 | 2 | Changelog |
3 | 3 | ********* |
4 | 4 |
|
5 | | -Version 1.5.0 (YYYY-MM-DD) |
| 5 | +Version 1.5.0 (2026-06-02) |
6 | 6 | ========================== |
7 | 7 |
|
| 8 | +**New features:** |
| 9 | + |
| 10 | +- Add new preprocessing methods for cleaning trajectory data before analysis: |
| 11 | + * Project trajectories into walkable area |
| 12 | + * Added functions for outlier and anomaly detection |
| 13 | +- Add method to compute RSET maps |
| 14 | + |
8 | 15 | **What's changed:** |
9 | 16 |
|
| 17 | +- Add support for Python 3.14 |
10 | 18 | - Improved Voronoi polygon computation with ~45% runtime improvement: |
11 | | - * Replaced manual scipy-based reconstruction with :code:`shapely.voronoi_polygons` |
12 | | - * Vectorized intersection and cutoff operations using shapely array operations |
| 19 | + * Behavioral change: :func:`~compute_individual_voronoi_polygons` now computes Voronoi cells for frames with fewer than 4 pedestrians. |
| 20 | + Previously, with :code:`use_blind_points=False`, such frames were skipped. The :code:`use_blind_points` parameter is now deprecated and has no effect. |
| 21 | + If you need to replicate the old behavior, filter the results manually: |
13 | 22 |
|
14 | | -**Migration guide:** |
| 23 | + .. code-block:: python |
15 | 24 |
|
16 | | -- **Behavioral change**: :func:`~compute_individual_voronoi_polygons` now computes Voronoi cells for frames with fewer than 4 pedestrians. |
17 | | -Previously, with :code:`use_blind_points=False`, such frames were skipped. The :code:`use_blind_points` parameter is now deprecated and has no effect. |
18 | | -If you need to replicate the old behavior, filter the results manually: |
| 25 | + # Filter out frames with < 4 pedestrians (old behavior) |
| 26 | + result = compute_individual_voronoi_polygons(traj_data, walkable_area) |
| 27 | + peds_per_frame = traj_data.data.groupby('frame').size() |
| 28 | + result = result[result['frame'].isin(peds_per_frame[peds_per_frame >= 4].index)] |
19 | 29 |
|
20 | | - .. code-block:: python |
| 30 | +- Improve profile computation memory efficiency |
| 31 | +- Unification of plot function options: All plotting functions now accept a unified set of keyword arguments, improving consistency and ease of use across different plot types. |
| 32 | + |
| 33 | +**Fixes:** |
21 | 34 |
|
22 | | - # Filter out frames with < 4 pedestrians (old behavior) |
23 | | - result = compute_individual_voronoi_polygons(traj_data, walkable_area) |
24 | | - peds_per_frame = traj_data.data.groupby('frame').size() |
25 | | - result = result[result['frame'].isin(peds_per_frame[peds_per_frame >= 4].index)] |
| 35 | +- Correctly plot Voronoi cells |
| 36 | +- Return Pandas DataFrame in all analysis functions as specified in the documentation |
| 37 | +- Minor plotting issues |
26 | 38 |
|
27 | 39 | Version 1.4.0 (2025-09-08) |
28 | 40 | ========================== |
|
0 commit comments