Skip to content

Commit

Permalink
updated numbers
Browse files Browse the repository at this point in the history
  • Loading branch information
@kprokofi
kprokofi committed May 2, 2024
1 parent ecab372 commit 360ccd0
Showing 1 changed file with 14 additions and 14 deletions.
28 changes: 14 additions & 14 deletions docs/source/guide/explanation/algorithms/segmentation/semantic_segmentation.rst
View file
Original file line number Diff line number Diff line change
Expand Up @@ -51,19 +51,19 @@ We support the following ready-to-use model recipes:
+--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+---------------------+-----------------+-----------------+-----------------+
| Recipe Path | Complexity (GFLOPs) | Model size (M) | FPS (GPU) | iter time (sec) |
+======================================================================================================================================================================================+=====================+=================+=================+=================+
| `Lite-HRNet-s-mod2 <https://github.com/openvinotoolkit/training_extensions/blob/develop/src/otx/recipe/semantic_segmentation/litehrnet_s.yaml>`_ | 1.44 | 3.2 | 37.68 | 0.151 |
| `Lite-HRNet-s-mod2 <https://github.com/openvinotoolkit/training_extensions/blob/develop/src/otx/recipe/semantic_segmentation/litehrnet_s.yaml>`_ | 1.44 | 0.82 | 37.68 | 0.151 |
+--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+---------------------+-----------------+-----------------+-----------------+
| `Lite-HRNet-18-mod2 <https://github.com/openvinotoolkit/training_extensions/blob/develop/src/otx/recipe/semantic_segmentation/litehrnet_18.yaml>`_ | 2.63 | 4.3 | 31.17 | 0.176 |
| `Lite-HRNet-18-mod2 <https://github.com/openvinotoolkit/training_extensions/blob/develop/src/otx/recipe/semantic_segmentation/litehrnet_18.yaml>`_ | 2.63 | 1.10 | 31.17 | 0.176 |
+--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+---------------------+-----------------+-----------------+-----------------+
| `Lite-HRNet-x-mod3 <https://github.com/openvinotoolkit/training_extensions/blob/develop/src/otx/recipe/semantic_segmentation/litehrnet_x.yaml>`_ | 9.20 | 5.7 | 15.07 | 0.347 |
| `Lite-HRNet-x-mod3 <https://github.com/openvinotoolkit/training_extensions/blob/develop/src/otx/recipe/semantic_segmentation/litehrnet_x.yaml>`_ | 9.20 | 1.50 | 15.07 | 0.347 |
+--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+---------------------+-----------------+-----------------+-----------------+
| `SegNext_T <https://github.com/openvinotoolkit/training_extensions/blob/develop/src/otx/recipe/semantic_segmentation/segnext_t.yaml>`_ | 6.07 | 4.23 | 104.90 | 0.126 |
| `SegNext_T <https://github.com/openvinotoolkit/training_extensions/blob/develop/src/otx/recipe/semantic_segmentation/segnext_t.yaml>`_ | 12.44 | 4.23 | 104.90 | 0.126 |
+--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+---------------------+-----------------+-----------------+-----------------+
| `SegNext_S <https://github.com/openvinotoolkit/training_extensions/blob/develop/src/otx/recipe/semantic_segmentation/segnext_s.yaml>`_ | 15.35 | 13.9 | 85.67 | 0.134 |
| `SegNext_S <https://github.com/openvinotoolkit/training_extensions/blob/develop/src/otx/recipe/semantic_segmentation/segnext_s.yaml>`_ | 30.93 | 13.90 | 85.67 | 0.134 |
+--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+---------------------+-----------------+-----------------+-----------------+
| `SegNext_B <https://github.com/openvinotoolkit/training_extensions/blob/develop/src/otx/recipe/semantic_segmentation/segnext_b.yaml>`_ | 32.08 | 27.56 | 61.91 | 0.215 |
| `SegNext_B <https://github.com/openvinotoolkit/training_extensions/blob/develop/src/otx/recipe/semantic_segmentation/segnext_b.yaml>`_ | 64.65 | 27.56 | 61.91 | 0.215 |
+--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+---------------------+-----------------+-----------------+-----------------+
| `DinoV2 <https://github.com/openvinotoolkit/training_extensions/blob/develop/src/otx/recipe/semantic_segmentation/dino_v2.yaml>`_ | 124.01 | 24.4 | 3.52 | 0.116 |
| `DinoV2 <https://github.com/openvinotoolkit/training_extensions/blob/develop/src/otx/recipe/semantic_segmentation/dino_v2.yaml>`_ | 124.01 | 24.40 | 3.52 | 0.116 |
+--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+---------------------+-----------------+-----------------+-----------------+

All of these models differ in the trade-off between accuracy and inference/training speed. For example, ``SegNext_B`` is the recipe with heavy-size architecture for more accurate predictions, but it requires longer training.
Expand All @@ -75,19 +75,19 @@ In the table below the `Dice score <https://en.wikipedia.org/wiki/S%C3%B8rensen%
+-----------------------+--------------------------------------------------------------+-----------------------------------------------------+----------------------------------------------------------------------+-----------------------------------------------------------------+--------+
| Model name | `DIS5K <https://xuebinqin.github.io/dis/index.html>`_ | `Cityscapes <https://www.cityscapes-dataset.com/>`_ | `Pascal-VOC 2012 <http://host.robots.ox.ac.uk/pascal/VOC/voc2012/>`_ | `KITTI <https://www.cvlibs.net/datasets/kitti/index.php>`_ | Mean |
+=======================+==============================================================+=====================================================+======================================================================+=================================================================+========+
| Lite-HRNet-s-mod2 | 79.95 | 62.38 | 63.26 | 41.73 | 59.16 |
| Lite-HRNet-s-mod2 | 78.73 | 69.25 | 63.26 | 41.73 | 59.16 |
+-----------------------+--------------------------------------------------------------+-----------------------------------------------------+----------------------------------------------------------------------+-----------------------------------------------------------------+--------+
| Lite-HRNet-18-mod2 | 81.12 | 65.04 | 62.10 | 46.73 | 62.20 |
| Lite-HRNet-18-mod2 | 81.43 | 72.66 | 62.10 | 46.73 | 62.20 |
+-----------------------+--------------------------------------------------------------+-----------------------------------------------------+----------------------------------------------------------------------+-----------------------------------------------------------------+--------+
| Lite-HRNet-x-mod3 | 79.98 | 59.97 | 59.55 | 49.97 | 60.85 |
| Lite-HRNet-x-mod3 | 82.36 | 74.57 | 59.55 | 49.97 | 60.85 |
+-----------------------+--------------------------------------------------------------+-----------------------------------------------------+----------------------------------------------------------------------+-----------------------------------------------------------------+--------+
| SegNext-t | 85.05 | 70.67 | 84.05 | 48.99 | 68.99 |
| SegNext-t | 83.99 | 77.09 | 84.05 | 48.99 | 68.99 |
+-----------------------+--------------------------------------------------------------+-----------------------------------------------------+----------------------------------------------------------------------+-----------------------------------------------------------------+--------+
| SegNext-s | 85.62 | 70.91 | 86.00 | 52.19 | 69.82 |
| SegNext-s | 85.54 | 79.45 | 86.00 | 52.19 | 69.82 |
+-----------------------+--------------------------------------------------------------+-----------------------------------------------------+----------------------------------------------------------------------+-----------------------------------------------------------------+--------+
| SegNext-b | 87.92 | 76.94 | 87.92 | 57.73 | 73.45 |
| SegNext-b | 86.76 | 76.14 | 87.92 | 57.73 | 73.45 |
+-----------------------+--------------------------------------------------------------+-----------------------------------------------------+----------------------------------------------------------------------+-----------------------------------------------------------------+--------+
| DinoV2 | xx.xx | xx.xx | xx.xx | xx.xx | 73.45 |
| DinoV2 | 84.87 | 73.58 | xx.xx | 65.91 | 73.45 |
+-----------------------+--------------------------------------------------------------+-----------------------------------------------------+----------------------------------------------------------------------+-----------------------------------------------------------------+--------+

.. note::
Expand Down

0 comments on commit 360ccd0

Please sign in to comment.