Convert dummy datasets to toy datasets (#1988)

* Update cls, det datsets * Remove useless files * Change action datasets * Edit action dataset * change dir * Add xml files * Remove useless * Edite tets * Fix tests * Fix tests * Remove ptc * Remove * Fix precommit * Update dataset, fix cls bug * Remove useless dataset * Edit drop_last * Fix missed part * Change threshold values to unifying * bugfix: squeezing to 1 dimenetion * Change threshold for deployment * Fix multi-gpu issue, e2e tests * Decrease num_workers for tiling test and tiling processes * Revert num_workers for tests * Fix datsets --------- Co-authored-by: eunwoosh <eunwoo.shin@intel.com>
openvinotoolkit · Apr 18, 2023 · ebb2d77 · ebb2d77
1 parent 861157a
commit ebb2d77
Show file tree

Hide file tree

Showing 217 changed files with 289 additions and 269 deletions.
diff --git a/otx/algorithms/classification/adapters/mmcls/configurer.py b/otx/algorithms/classification/adapters/mmcls/configurer.py
@@ -426,7 +426,7 @@ def configure_samples_per_gpu(
             # drop the last batch if the last batch size is 1
             # batch size of 1 is a runtime error for training batch normalization layer
             if subset in ("train", "unlabeled") and dataset_len % samples_per_gpu == 1:
-                dataloader_cfg.drop_last = True
+                dataloader_cfg["drop_last"] = True
 
             cfg.data[f"{subset}_dataloader"] = dataloader_cfg
 

diff --git a/otx/algorithms/classification/adapters/mmcls/models/classifiers/semisl_classifier.py b/otx/algorithms/classification/adapters/mmcls/models/classifiers/semisl_classifier.py
@@ -32,7 +32,7 @@ def forward_train(self, imgs, **kwargs):
             raise ValueError("'gt_label' does not exist in the labeled image")
         if "extra_0" not in kwargs:
             raise ValueError("'extra_0' does not exist in the dataset")
-        target = kwargs["gt_label"].squeeze()
+        target = kwargs["gt_label"].squeeze(dim=1)
         unlabeled_data = kwargs["extra_0"]
         x = {}
         x["labeled"] = self.extract_feat(imgs)

diff --git a/otx/algorithms/classification/adapters/mmcls/utils/config_utils.py b/otx/algorithms/classification/adapters/mmcls/utils/config_utils.py
@@ -55,7 +55,7 @@ def update_pipeline(cfg):
 
         # For stable hierarchical information indexing
         if subset == "train" and kwargs["type"] == "OTXHierarchicalClsDataset":
-            config.data[f"{subset}_dataloader"].drop_last = True
+            config.data[f"{subset}_dataloader"]["drop_last"] = True
 
         cfgs = get_dataset_configs(config, subset)
         for cfg in cfgs:

diff --git a/otx/algorithms/detection/adapters/mmdet/configurer.py b/otx/algorithms/detection/adapters/mmdet/configurer.py
@@ -588,7 +588,7 @@ def configure_samples_per_gpu(
             # drop the last batch if the last batch size is 1
             # batch size of 1 is a runtime error for training batch normalization layer
             if subset in ("train", "unlabeled") and dataset_len % samples_per_gpu == 1:
-                dataloader_cfg.drop_last = True
+                dataloader_cfg["drop_last"] = True
 
             cfg.data[f"{subset}_dataloader"] = dataloader_cfg
 

diff --git a/otx/algorithms/detection/adapters/mmdet/datasets/tiling.py b/otx/algorithms/detection/adapters/mmdet/datasets/tiling.py
@@ -74,7 +74,7 @@ def __init__(
         iou_threshold: float = 0.45,
         max_per_img: int = 1500,
         filter_empty_gt: bool = True,
-        nproc: int = 4,
+        nproc: int = 2,
     ):
         self.min_area_ratio = min_area_ratio
         self.filter_empty_gt = filter_empty_gt

diff --git a/otx/algorithms/segmentation/adapters/mmseg/configurer.py b/otx/algorithms/segmentation/adapters/mmseg/configurer.py
@@ -420,7 +420,7 @@ def configure_samples_per_gpu(self, cfg: Config, subset: str) -> None:
             # drop the last batch if the last batch size is 1
             # batch size of 1 is a runtime error for training batch normalization layer
             if subset in ("train", "unlabeled") and dataset_len % samples_per_gpu == 1:
-                dataloader_cfg.drop_last = True
+                dataloader_cfg["drop_last"] = True
 
             cfg.data[f"{subset}_dataloader"] = dataloader_cfg
 

diff --git a/otx/cli/utils/multi_gpu.py b/otx/cli/utils/multi_gpu.py
@@ -191,7 +191,7 @@ def setup_multi_gpu_train(
     def finalize(self):
         """Join all child processes."""
         for p in self._processes:
-            if p.join(10) is None and p.exitcode is None:
+            if p.join(30) is None and p.exitcode is None:
                 p.kill()
 
     @staticmethod

diff --git a/tests/assets/classification_dataset/0/11.jpg b/tests/assets/classification_dataset/0/11.jpg
diff --git a/tests/assets/classification_dataset/0/14.jpg b/tests/assets/classification_dataset/0/14.jpg
diff --git a/tests/assets/classification_dataset/0/16.jpg b/tests/assets/classification_dataset/0/16.jpg
diff --git a/tests/assets/classification_dataset/0/17.jpg b/tests/assets/classification_dataset/0/17.jpg
diff --git a/tests/assets/classification_dataset/0/18.jpg b/tests/assets/classification_dataset/0/18.jpg
diff --git a/...n_classification/train/2/images/00022.jpg → tests/assets/classification_dataset/0/23.jpg b/...n_classification/train/2/images/00022.jpg → tests/assets/classification_dataset/0/23.jpg
diff --git a/tests/assets/classification_dataset/0/26.jpg b/tests/assets/classification_dataset/0/26.jpg
diff --git a/tests/assets/classification_dataset/0/3.jpg b/tests/assets/classification_dataset/0/3.jpg
diff --git a/tests/assets/classification_dataset/0/30.jpg b/tests/assets/classification_dataset/0/30.jpg
diff --git a/tests/assets/classification_dataset/0/4.jpg b/tests/assets/classification_dataset/0/4.jpg
diff --git a/tests/assets/classification_dataset/0/5.jpg b/tests/assets/classification_dataset/0/5.jpg
diff --git a/tests/assets/classification_dataset/0/6.jpg b/tests/assets/classification_dataset/0/6.jpg
diff --git a/...n_classification/train/2/images/00023.jpg → tests/assets/classification_dataset/0/7.jpg b/...n_classification/train/2/images/00023.jpg → tests/assets/classification_dataset/0/7.jpg
diff --git a/tests/assets/classification_dataset/0/8.jpg b/tests/assets/classification_dataset/0/8.jpg
diff --git a/tests/assets/classification_dataset/1/0.jpg b/tests/assets/classification_dataset/1/0.jpg
diff --git a/tests/assets/classification_dataset/1/1.jpg b/tests/assets/classification_dataset/1/1.jpg
diff --git a/tests/assets/classification_dataset/1/10.jpg b/tests/assets/classification_dataset/1/10.jpg
diff --git a/tests/assets/classification_dataset/1/12.jpg b/tests/assets/classification_dataset/1/12.jpg
diff --git a/tests/assets/classification_dataset/1/15.jpg b/tests/assets/classification_dataset/1/15.jpg
diff --git a/tests/assets/classification_dataset/1/19.jpg b/tests/assets/classification_dataset/1/19.jpg
diff --git a/tests/assets/classification_dataset/1/2.jpg b/tests/assets/classification_dataset/1/2.jpg
diff --git a/tests/assets/classification_dataset/1/21.jpg b/tests/assets/classification_dataset/1/21.jpg
diff --git a/tests/assets/classification_dataset/1/27.jpg b/tests/assets/classification_dataset/1/27.jpg
diff --git a/tests/assets/classification_dataset/1/28.jpg b/tests/assets/classification_dataset/1/28.jpg
diff --git a/tests/assets/classification_dataset/1/29.jpg b/tests/assets/classification_dataset/1/29.jpg
diff --git a/tests/assets/classification_dataset_class_incremental/0/11.jpg b/tests/assets/classification_dataset_class_incremental/0/11.jpg
diff --git a/tests/assets/classification_dataset_class_incremental/0/14.jpg b/tests/assets/classification_dataset_class_incremental/0/14.jpg
diff --git a/tests/assets/classification_dataset_class_incremental/0/16.jpg b/tests/assets/classification_dataset_class_incremental/0/16.jpg
diff --git a/tests/assets/classification_dataset_class_incremental/0/17.jpg b/tests/assets/classification_dataset_class_incremental/0/17.jpg
diff --git a/tests/assets/classification_dataset_class_incremental/0/18.jpg b/tests/assets/classification_dataset_class_incremental/0/18.jpg
diff --git a/tests/assets/classification_dataset_class_incremental/0/23.jpg b/tests/assets/classification_dataset_class_incremental/0/23.jpg
diff --git a/tests/assets/classification_dataset_class_incremental/0/26.jpg b/tests/assets/classification_dataset_class_incremental/0/26.jpg
diff --git a/tests/assets/classification_dataset_class_incremental/0/3.jpg b/tests/assets/classification_dataset_class_incremental/0/3.jpg
diff --git a/tests/assets/classification_dataset_class_incremental/0/30.jpg b/tests/assets/classification_dataset_class_incremental/0/30.jpg
diff --git a/tests/assets/classification_dataset_class_incremental/0/4.jpg b/tests/assets/classification_dataset_class_incremental/0/4.jpg
diff --git a/tests/assets/classification_dataset_class_incremental/0/5.jpg b/tests/assets/classification_dataset_class_incremental/0/5.jpg
diff --git a/tests/assets/classification_dataset_class_incremental/0/6.jpg b/tests/assets/classification_dataset_class_incremental/0/6.jpg
diff --git a/...n_classification/train/2/images/00020.jpg → ...ication_dataset_class_incremental/0/7.jpg b/...n_classification/train/2/images/00020.jpg → ...ication_dataset_class_incremental/0/7.jpg
diff --git a/tests/assets/classification_dataset_class_incremental/0/8.jpg b/tests/assets/classification_dataset_class_incremental/0/8.jpg
diff --git a/tests/assets/classification_dataset_class_incremental/1/0.jpg b/tests/assets/classification_dataset_class_incremental/1/0.jpg
diff --git a/tests/assets/classification_dataset_class_incremental/1/1.jpg b/tests/assets/classification_dataset_class_incremental/1/1.jpg
diff --git a/tests/assets/classification_dataset_class_incremental/1/10.jpg b/tests/assets/classification_dataset_class_incremental/1/10.jpg
diff --git a/tests/assets/classification_dataset_class_incremental/1/12.jpg b/tests/assets/classification_dataset_class_incremental/1/12.jpg
diff --git a/tests/assets/classification_dataset_class_incremental/1/15.jpg b/tests/assets/classification_dataset_class_incremental/1/15.jpg
diff --git a/tests/assets/classification_dataset_class_incremental/1/19.jpg b/tests/assets/classification_dataset_class_incremental/1/19.jpg
diff --git a/tests/assets/classification_dataset_class_incremental/1/2.jpg b/tests/assets/classification_dataset_class_incremental/1/2.jpg
diff --git a/tests/assets/classification_dataset_class_incremental/1/21.jpg b/tests/assets/classification_dataset_class_incremental/1/21.jpg
diff --git a/tests/assets/classification_dataset_class_incremental/1/27.jpg b/tests/assets/classification_dataset_class_incremental/1/27.jpg
diff --git a/tests/assets/classification_dataset_class_incremental/1/28.jpg b/tests/assets/classification_dataset_class_incremental/1/28.jpg
diff --git a/tests/assets/classification_dataset_class_incremental/1/29.jpg b/tests/assets/classification_dataset_class_incremental/1/29.jpg
diff --git a/...n_classification/train/2/images/00021.jpg → ...cation_dataset_class_incremental/2/13.jpg b/...n_classification/train/2/images/00021.jpg → ...cation_dataset_class_incremental/2/13.jpg
diff --git a/tests/assets/classification_dataset_class_incremental/2/20.jpg b/tests/assets/classification_dataset_class_incremental/2/20.jpg
diff --git a/tests/assets/classification_dataset_class_incremental/2/22.jpg b/tests/assets/classification_dataset_class_incremental/2/22.jpg
diff --git a/tests/assets/classification_dataset_class_incremental/2/24.jpg b/tests/assets/classification_dataset_class_incremental/2/24.jpg
diff --git a/tests/assets/classification_dataset_class_incremental/2/25.jpg b/tests/assets/classification_dataset_class_incremental/2/25.jpg
diff --git a/tests/assets/classification_dataset_class_incremental/2/31.jpg b/tests/assets/classification_dataset_class_incremental/2/31.jpg
diff --git a/tests/assets/classification_dataset_class_incremental/2/9.jpg b/tests/assets/classification_dataset_class_incremental/2/9.jpg
diff --git a/tests/assets/cvat_dataset/action_classification/train/0/annotations.xml b/tests/assets/cvat_dataset/action_classification/train/0/annotations.xml
@@ -23,39 +23,39 @@
       </labels>
       <segments></segments>
       <original_size>
-        <width>256</width>
-        <height>192</height>
+        <width>24</width>
+        <height>24</height>
       </original_size>
     </task>
   </meta>
-  <image id="0" name="00020.jpg" width="256" height="192">
+  <image id="0" name="00020.jpg" width="24" height="24">
     <tag label="0" source="manual"></tag>
   </image>
-  <image id="1" name="00021.jpg" width="256" height="192">
+  <image id="1" name="00021.jpg" width="24" height="24">
     <tag label="0" source="manual"></tag>
   </image>
-  <image id="2" name="00022.jpg" width="256" height="192">
+  <image id="2" name="00022.jpg" width="24" height="24">
     <tag label="0" source="manual"></tag>
   </image>
-  <image id="3" name="00023.jpg" width="256" height="192">
+  <image id="3" name="00023.jpg" width="24" height="24">
     <tag label="0" source="manual"></tag>
   </image>
-  <image id="4" name="00024.jpg" width="256" height="192">
+  <image id="4" name="00024.jpg" width="24" height="24">
     <tag label="0" source="manual"></tag>
   </image>
-  <image id="5" name="00025.jpg" width="256" height="192">
+  <image id="5" name="00025.jpg" width="24" height="24">
     <tag label="0" source="manual"></tag>
   </image>
-  <image id="6" name="00026.jpg" width="256" height="192">
+  <image id="6" name="00026.jpg" width="24" height="24">
     <tag label="0" source="manual"></tag>
   </image>
-  <image id="7" name="00027.jpg" width="256" height="192">
+  <image id="7" name="00027.jpg" width="24" height="24">
     <tag label="0" source="manual"></tag>
   </image>
-  <image id="8" name="00028.jpg" width="256" height="192">
+  <image id="8" name="00028.jpg" width="24" height="24">
     <tag label="0" source="manual"></tag>
   </image>
-  <image id="9" name="00029.jpg" width="256" height="192">
+  <image id="9" name="00029.jpg" width="24" height="24">
     <tag label="0" source="manual"></tag>
   </image>
 </annotations>
diff --git a/tests/assets/cvat_dataset/action_classification/train/0/images/00020.jpg b/tests/assets/cvat_dataset/action_classification/train/0/images/00020.jpg
diff --git a/tests/assets/cvat_dataset/action_classification/train/0/images/00021.jpg b/tests/assets/cvat_dataset/action_classification/train/0/images/00021.jpg
diff --git a/tests/assets/cvat_dataset/action_classification/train/0/images/00022.jpg b/tests/assets/cvat_dataset/action_classification/train/0/images/00022.jpg
diff --git a/tests/assets/cvat_dataset/action_classification/train/0/images/00023.jpg b/tests/assets/cvat_dataset/action_classification/train/0/images/00023.jpg
diff --git a/tests/assets/cvat_dataset/action_classification/train/0/images/00024.jpg b/tests/assets/cvat_dataset/action_classification/train/0/images/00024.jpg
diff --git a/tests/assets/cvat_dataset/action_classification/train/0/images/00025.jpg b/tests/assets/cvat_dataset/action_classification/train/0/images/00025.jpg
diff --git a/tests/assets/cvat_dataset/action_classification/train/0/images/00026.jpg b/tests/assets/cvat_dataset/action_classification/train/0/images/00026.jpg
diff --git a/tests/assets/cvat_dataset/action_classification/train/0/images/00027.jpg b/tests/assets/cvat_dataset/action_classification/train/0/images/00027.jpg
diff --git a/tests/assets/cvat_dataset/action_classification/train/0/images/00028.jpg b/tests/assets/cvat_dataset/action_classification/train/0/images/00028.jpg
diff --git a/tests/assets/cvat_dataset/action_classification/train/0/images/00029.jpg b/tests/assets/cvat_dataset/action_classification/train/0/images/00029.jpg
diff --git a/tests/assets/cvat_dataset/action_classification/train/1/annotations.xml b/tests/assets/cvat_dataset/action_classification/train/1/annotations.xml
@@ -3,8 +3,8 @@
   <version>1.1</version>
   <meta>
     <task>
-      <id>2</id>
-      <name>v2</name>
+      <id>1</id>
+      <name>v1</name>
       <size>10</size>
       <mode>interpolation</mode>
       <overlap>2</overlap>
@@ -23,39 +23,39 @@
       </labels>
       <segments></segments>
       <original_size>
-        <width>256</width>
-        <height>192</height>
+        <width>24</width>
+        <height>24</height>
       </original_size>
     </task>
   </meta>
-  <image id="0" name="00020.jpg" width="256" height="192">
+  <image id="0" name="00020.jpg" width="24" height="24">
     <tag label="1" source="manual"></tag>
   </image>
-  <image id="1" name="00021.jpg" width="256" height="192">
+  <image id="1" name="00021.jpg" width="24" height="24">
     <tag label="1" source="manual"></tag>
   </image>
-  <image id="2" name="00022.jpg" width="256" height="192">
+  <image id="2" name="00022.jpg" width="24" height="24">
     <tag label="1" source="manual"></tag>
   </image>
-  <image id="3" name="00023.jpg" width="256" height="192">
+  <image id="3" name="00023.jpg" width="24" height="24">
     <tag label="1" source="manual"></tag>
   </image>
-  <image id="4" name="00024.jpg" width="256" height="192">
+  <image id="4" name="00024.jpg" width="24" height="24">
     <tag label="1" source="manual"></tag>
   </image>
-  <image id="5" name="00025.jpg" width="256" height="192">
+  <image id="5" name="00025.jpg" width="24" height="24">
     <tag label="1" source="manual"></tag>
   </image>
-  <image id="6" name="00026.jpg" width="256" height="192">
+  <image id="6" name="00026.jpg" width="24" height="24">
     <tag label="1" source="manual"></tag>
   </image>
-  <image id="7" name="00027.jpg" width="256" height="192">
+  <image id="7" name="00027.jpg" width="24" height="24">
     <tag label="1" source="manual"></tag>
   </image>
-  <image id="8" name="00028.jpg" width="256" height="192">
+  <image id="8" name="00028.jpg" width="24" height="24">
     <tag label="1" source="manual"></tag>
   </image>
-  <image id="9" name="00029.jpg" width="256" height="192">
+  <image id="9" name="00029.jpg" width="24" height="24">
     <tag label="1" source="manual"></tag>
   </image>
 </annotations>
diff --git a/tests/assets/cvat_dataset/action_classification/train/1/images/00020.jpg b/tests/assets/cvat_dataset/action_classification/train/1/images/00020.jpg
diff --git a/tests/assets/cvat_dataset/action_classification/train/1/images/00021.jpg b/tests/assets/cvat_dataset/action_classification/train/1/images/00021.jpg
diff --git a/tests/assets/cvat_dataset/action_classification/train/1/images/00022.jpg b/tests/assets/cvat_dataset/action_classification/train/1/images/00022.jpg
diff --git a/tests/assets/cvat_dataset/action_classification/train/1/images/00023.jpg b/tests/assets/cvat_dataset/action_classification/train/1/images/00023.jpg
diff --git a/tests/assets/cvat_dataset/action_classification/train/1/images/00024.jpg b/tests/assets/cvat_dataset/action_classification/train/1/images/00024.jpg
diff --git a/tests/assets/cvat_dataset/action_classification/train/1/images/00025.jpg b/tests/assets/cvat_dataset/action_classification/train/1/images/00025.jpg
diff --git a/tests/assets/cvat_dataset/action_classification/train/1/images/00026.jpg b/tests/assets/cvat_dataset/action_classification/train/1/images/00026.jpg
diff --git a/tests/assets/cvat_dataset/action_classification/train/1/images/00027.jpg b/tests/assets/cvat_dataset/action_classification/train/1/images/00027.jpg
diff --git a/tests/assets/cvat_dataset/action_classification/train/1/images/00028.jpg b/tests/assets/cvat_dataset/action_classification/train/1/images/00028.jpg
diff --git a/tests/assets/cvat_dataset/action_classification/train/1/images/00029.jpg b/tests/assets/cvat_dataset/action_classification/train/1/images/00029.jpg
diff --git a/tests/assets/cvat_dataset/action_classification/train/2/annotations.xml b/tests/assets/cvat_dataset/action_classification/train/2/annotations.xml
diff --git a/tests/assets/cvat_dataset/action_classification/train/2/images/00024.jpg b/tests/assets/cvat_dataset/action_classification/train/2/images/00024.jpg
diff --git a/tests/assets/cvat_dataset/action_classification/train/2/images/00025.jpg b/tests/assets/cvat_dataset/action_classification/train/2/images/00025.jpg
diff --git a/tests/assets/cvat_dataset/action_classification/train/2/images/00026.jpg b/tests/assets/cvat_dataset/action_classification/train/2/images/00026.jpg
diff --git a/tests/assets/cvat_dataset/action_classification/train/2/images/00027.jpg b/tests/assets/cvat_dataset/action_classification/train/2/images/00027.jpg
diff --git a/tests/assets/cvat_dataset/action_classification/train/2/images/00028.jpg b/tests/assets/cvat_dataset/action_classification/train/2/images/00028.jpg
diff --git a/tests/assets/cvat_dataset/action_classification/train/2/images/00029.jpg b/tests/assets/cvat_dataset/action_classification/train/2/images/00029.jpg
diff --git a/tests/assets/cvat_dataset/action_detection/train.pkl b/tests/assets/cvat_dataset/action_detection/train.pkl
diff --git a/tests/assets/cvat_dataset/action_detection/train/0/annotations.xml b/tests/assets/cvat_dataset/action_detection/train/0/annotations.xml
@@ -21,45 +21,45 @@
           <attributes></attributes>
         </label>
         <label>
-          <name>1</name>
+          <name>Right</name>
           <attributes></attributes>
         </label>
       </labels>
       <segments></segments>
       <original_size>
-        <width>256</width>
-        <height>192</height>
+        <width>24</width>
+        <height>24</height>
       </original_size>
     </task>
   </meta>
-  <track id="0" label="EmptyFrame">
-    <box frame="20" outside="0" occluded="0" xtl="0" ytl="0" xbr="0" ybr="0" z_order="0" label="EmptyFrame"></box>
+  <track id="0" label="Right">
+    <box frame="00" outside="0" occluded="0" xtl="0.08" ytl="0.25" xbr="0.50" ybr="0.75" z_order="0" label="Right"></box>
   </track>
-  <track id="0" label="EmptyFrame">
-    <box frame="21" outside="0" occluded="0" xtl="0" ytl="0" xbr="0" ybr="0" z_order="0" label="EmptyFrame"></box>
+  <track id="0" label="Right">
+    <box frame="01" outside="0" occluded="0" xtl="0.12" ytl="0.25" xbr="0.54" ybr="0.75" z_order="0" label="Right"></box>
   </track>
-  <track id="0" label="EmptyFrame">
-    <box frame="22" outside="0" occluded="0" xtl="0" ytl="0" xbr="0" ybr="0" z_order="0" label="EmptyFrame"></box>
+  <track id="0" label="Right">
+    <box frame="02" outside="0" occluded="0" xtl="0.16" ytl="0.25" xbr="0.58" ybr="0.75" z_order="0" label="Right"></box>
   </track>
-  <track id="0" label="EmptyFrame">
-    <box frame="23" outside="0" occluded="0" xtl="0" ytl="0" xbr="0" ybr="0" z_order="0" label="EmptyFrame"></box>
+  <track id="0" label="Right">
+    <box frame="03" outside="0" occluded="0" xtl="0.20" ytl="0.25" xbr="0.62" ybr="0.75" z_order="0" label="Right"></box>
   </track>
-  <track id="0" label="1">
-    <box frame="24" outside="0" occluded="0" xtl="0.27298716" ytl="0.0073581696" xbr="0.67030156" ybr="0.9858064" z_order="0" label="1"></box>
+  <track id="0" label="Right">
+    <box frame="04" outside="0" occluded="0" xtl="0.25" ytl="0.25" xbr="0.67" ybr="0.75" z_order="0" label="Right"></box>
   </track>
-  <track id="0" label="EmptyFrame">
-    <box frame="25" outside="0" occluded="0" xtl="0" ytl="0" xbr="0" ybr="0" z_order="0" label="EmptyFrame"></box>
+  <track id="0" label="Right">
+    <box frame="05" outside="0" occluded="0" xtl="0.29" ytl="0.25" xbr="0.71" ybr="0.75" z_order="0" label="Right"></box>
   </track>
-  <track id="0" label="EmptyFrame">
-    <box frame="26" outside="0" occluded="0" xtl="0" ytl="0" xbr="0" ybr="0" z_order="0" label="EmptyFrame"></box>
+  <track id="0" label="Right">
+    <box frame="06" outside="0" occluded="0" xtl="0.33" ytl="0.25" xbr="0.75" ybr="0.75" z_order="0" label="Right"></box>
   </track>
-  <track id="0" label="EmptyFrame">
-    <box frame="27" outside="0" occluded="0" xtl="0" ytl="0" xbr="0" ybr="0" z_order="0" label="EmptyFrame"></box>
+  <track id="0" label="Right">
+    <box frame="07" outside="0" occluded="0" xtl="0.37" ytl="0.25" xbr="0.79" ybr="0.75" z_order="0" label="Right"></box>
   </track>
-  <track id="0" label="EmptyFrame">
-    <box frame="28" outside="0" occluded="0" xtl="0" ytl="0" xbr="0" ybr="0" z_order="0" label="EmptyFrame"></box>
+  <track id="0" label="Right">
+    <box frame="08" outside="0" occluded="0" xtl="0.41" ytl="0.25" xbr="0.83" ybr="0.75" z_order="0" label="Right"></box>
   </track>
-  <track id="0" label="EmptyFrame">
-    <box frame="29" outside="0" occluded="0" xtl="0" ytl="0" xbr="0" ybr="0" z_order="0" label="EmptyFrame"></box>
+  <track id="0" label="Right">
+    <box frame="09" outside="0" occluded="0" xtl="0.45" ytl="0.25" xbr="0.87" ybr="0.75" z_order="0" label="Right"></box>
   </track>
 </annotations>
diff --git a/tests/assets/cvat_dataset/action_detection/train/0/images/frame_000000.png b/tests/assets/cvat_dataset/action_detection/train/0/images/frame_000000.png
diff --git a/tests/assets/cvat_dataset/action_detection/train/0/images/frame_000001.png b/tests/assets/cvat_dataset/action_detection/train/0/images/frame_000001.png
diff --git a/tests/assets/cvat_dataset/action_detection/train/0/images/frame_000002.png b/tests/assets/cvat_dataset/action_detection/train/0/images/frame_000002.png
diff --git a/tests/assets/cvat_dataset/action_detection/train/0/images/frame_000003.png b/tests/assets/cvat_dataset/action_detection/train/0/images/frame_000003.png
diff --git a/tests/assets/cvat_dataset/action_detection/train/0/images/frame_000004.png b/tests/assets/cvat_dataset/action_detection/train/0/images/frame_000004.png
diff --git a/tests/assets/cvat_dataset/action_detection/train/0/images/frame_000005.png b/tests/assets/cvat_dataset/action_detection/train/0/images/frame_000005.png
diff --git a/tests/assets/cvat_dataset/action_detection/train/0/images/frame_000006.png b/tests/assets/cvat_dataset/action_detection/train/0/images/frame_000006.png
diff --git a/tests/assets/cvat_dataset/action_detection/train/0/images/frame_000007.png b/tests/assets/cvat_dataset/action_detection/train/0/images/frame_000007.png
diff --git a/tests/assets/cvat_dataset/action_detection/train/0/images/frame_000008.png b/tests/assets/cvat_dataset/action_detection/train/0/images/frame_000008.png
diff --git a/tests/assets/cvat_dataset/action_detection/train/0/images/frame_000009.png b/tests/assets/cvat_dataset/action_detection/train/0/images/frame_000009.png
diff --git a/tests/assets/cvat_dataset/action_detection/train/0/images/frame_000020.png b/tests/assets/cvat_dataset/action_detection/train/0/images/frame_000020.png
diff --git a/tests/assets/cvat_dataset/action_detection/train/0/images/frame_000021.png b/tests/assets/cvat_dataset/action_detection/train/0/images/frame_000021.png
diff --git a/tests/assets/cvat_dataset/action_detection/train/0/images/frame_000022.png b/tests/assets/cvat_dataset/action_detection/train/0/images/frame_000022.png
diff --git a/tests/assets/cvat_dataset/action_detection/train/0/images/frame_000023.png b/tests/assets/cvat_dataset/action_detection/train/0/images/frame_000023.png
diff --git a/tests/assets/cvat_dataset/action_detection/train/0/images/frame_000024.png b/tests/assets/cvat_dataset/action_detection/train/0/images/frame_000024.png
diff --git a/tests/assets/cvat_dataset/action_detection/train/0/images/frame_000025.png b/tests/assets/cvat_dataset/action_detection/train/0/images/frame_000025.png
diff --git a/tests/assets/cvat_dataset/action_detection/train/0/images/frame_000026.png b/tests/assets/cvat_dataset/action_detection/train/0/images/frame_000026.png
diff --git a/tests/assets/cvat_dataset/action_detection/train/0/images/frame_000027.png b/tests/assets/cvat_dataset/action_detection/train/0/images/frame_000027.png
diff --git a/tests/assets/cvat_dataset/action_detection/train/0/images/frame_000028.png b/tests/assets/cvat_dataset/action_detection/train/0/images/frame_000028.png
diff --git a/tests/assets/cvat_dataset/action_detection/train/0/images/frame_000029.png b/tests/assets/cvat_dataset/action_detection/train/0/images/frame_000029.png
diff --git a/tests/assets/cvat_dataset/action_detection/train/1/annotations.xml b/tests/assets/cvat_dataset/action_detection/train/1/annotations.xml
@@ -21,45 +21,45 @@
           <attributes></attributes>
         </label>
         <label>
-          <name>2</name>
+          <name>Left</name>
           <attributes></attributes>
         </label>
       </labels>
       <segments></segments>
       <original_size>
-        <width>256</width>
-        <height>192</height>
+        <width>24</width>
+        <height>24</height>
       </original_size>
     </task>
   </meta>
-  <track id="0" label="EmptyFrame">
-    <box frame="20" outside="0" occluded="0" xtl="0" ytl="0" xbr="0" ybr="0" z_order="0" label="EmptyFrame"></box>
+  <track id="0" label="Left">
+    <box frame="00" outside="0" occluded="0" xtl="0.45" ytl="0.25" xbr="0.87" ybr="0.75" z_order="0" label="Left"></box>
   </track>
-  <track id="0" label="EmptyFrame">
-    <box frame="21" outside="0" occluded="0" xtl="0" ytl="0" xbr="0" ybr="0" z_order="0" label="EmptyFrame"></box>
+  <track id="0" label="Left">
+    <box frame="01" outside="0" occluded="0" xtl="0.41" ytl="0.25" xbr="0.83" ybr="0.75" z_order="0" label="Left"></box>
   </track>
-  <track id="0" label="EmptyFrame">
-    <box frame="22" outside="0" occluded="0" xtl="0" ytl="0" xbr="0" ybr="0" z_order="0" label="EmptyFrame"></box>
+  <track id="0" label="Left">
+    <box frame="02" outside="0" occluded="0" xtl="0.37" ytl="0.25" xbr="0.79" ybr="0.75" z_order="0" label="Left"></box>
   </track>
-  <track id="0" label="2">
-    <box frame="23" outside="0" occluded="0" xtl="0.22982445" ytl="0.0969286" xbr="0.72086316" ybr="0.9888343" z_order="0" label="2"></box>
+  <track id="0" label="Left">
+    <box frame="03" outside="0" occluded="0" xtl="0.33" ytl="0.25" xbr="0.75" ybr="0.75" z_order="0" label="Left"></box>
   </track>
-  <track id="0" label="EmptyFrame">
-    <box frame="24" outside="0" occluded="0" xtl="0" ytl="0" xbr="0" ybr="0" z_order="0" label="EmptyFrame"></box>
+  <track id="0" label="Left">
+    <box frame="04" outside="0" occluded="0" xtl="0.29" ytl="0.25" xbr="0.71" ybr="0.75" z_order="0" label="Left"></box>
   </track>
-  <track id="0" label="EmptyFrame">
-    <box frame="25" outside="0" occluded="0" xtl="0" ytl="0" xbr="0" ybr="0" z_order="0" label="EmptyFrame"></box>
+  <track id="0" label="Left">
+    <box frame="05" outside="0" occluded="0" xtl="0.25" ytl="0.25" xbr="0.67" ybr="0.75" z_order="0" label="Left"></box>
   </track>
-  <track id="0" label="EmptyFrame">
-    <box frame="26" outside="0" occluded="0" xtl="0" ytl="0" xbr="0" ybr="0" z_order="0" label="EmptyFrame"></box>
+  <track id="0" label="Left">
+    <box frame="06" outside="0" occluded="0" xtl="0.20" ytl="0.25" xbr="0.62" ybr="0.75" z_order="0" label="Left"></box>
   </track>
-  <track id="0" label="EmptyFrame">
-    <box frame="27" outside="0" occluded="0" xtl="0" ytl="0" xbr="0" ybr="0" z_order="0" label="EmptyFrame"></box>
+  <track id="0" label="Left">
+    <box frame="07" outside="0" occluded="0" xtl="0.16" ytl="0.25" xbr="0.58" ybr="0.75" z_order="0" label="Left"></box>
   </track>
-  <track id="0" label="EmptyFrame">
-    <box frame="28" outside="0" occluded="0" xtl="0" ytl="0" xbr="0" ybr="0" z_order="0" label="EmptyFrame"></box>
+  <track id="0" label="Left">
+    <box frame="08" outside="0" occluded="0" xtl="0.12" ytl="0.25" xbr="0.54" ybr="0.75" z_order="0" label="Left"></box>
   </track>
-  <track id="0" label="EmptyFrame">
-    <box frame="29" outside="0" occluded="0" xtl="0" ytl="0" xbr="0" ybr="0" z_order="0" label="EmptyFrame"></box>
+  <track id="0" label="Left">
+    <box frame="09" outside="0" occluded="0" xtl="0.08" ytl="0.25" xbr="0.50" ybr="0.75" z_order="0" label="Left"></box>
   </track>
 </annotations>
diff --git a/tests/assets/cvat_dataset/action_detection/train/1/images/frame_000000.png b/tests/assets/cvat_dataset/action_detection/train/1/images/frame_000000.png
diff --git a/tests/assets/cvat_dataset/action_detection/train/1/images/frame_000001.png b/tests/assets/cvat_dataset/action_detection/train/1/images/frame_000001.png
diff --git a/tests/assets/cvat_dataset/action_detection/train/1/images/frame_000002.png b/tests/assets/cvat_dataset/action_detection/train/1/images/frame_000002.png
diff --git a/tests/assets/cvat_dataset/action_detection/train/1/images/frame_000003.png b/tests/assets/cvat_dataset/action_detection/train/1/images/frame_000003.png
diff --git a/tests/assets/cvat_dataset/action_detection/train/1/images/frame_000004.png b/tests/assets/cvat_dataset/action_detection/train/1/images/frame_000004.png
diff --git a/tests/assets/cvat_dataset/action_detection/train/1/images/frame_000005.png b/tests/assets/cvat_dataset/action_detection/train/1/images/frame_000005.png
diff --git a/tests/assets/cvat_dataset/action_detection/train/1/images/frame_000006.png b/tests/assets/cvat_dataset/action_detection/train/1/images/frame_000006.png
diff --git a/tests/assets/cvat_dataset/action_detection/train/1/images/frame_000007.png b/tests/assets/cvat_dataset/action_detection/train/1/images/frame_000007.png
diff --git a/tests/assets/cvat_dataset/action_detection/train/1/images/frame_000008.png b/tests/assets/cvat_dataset/action_detection/train/1/images/frame_000008.png
diff --git a/tests/assets/cvat_dataset/action_detection/train/1/images/frame_000009.png b/tests/assets/cvat_dataset/action_detection/train/1/images/frame_000009.png
diff --git a/tests/assets/cvat_dataset/action_detection/train/1/images/frame_000020.png b/tests/assets/cvat_dataset/action_detection/train/1/images/frame_000020.png
diff --git a/tests/assets/cvat_dataset/action_detection/train/1/images/frame_000021.png b/tests/assets/cvat_dataset/action_detection/train/1/images/frame_000021.png
diff --git a/tests/assets/cvat_dataset/action_detection/train/1/images/frame_000022.png b/tests/assets/cvat_dataset/action_detection/train/1/images/frame_000022.png
diff --git a/tests/assets/cvat_dataset/action_detection/train/1/images/frame_000023.png b/tests/assets/cvat_dataset/action_detection/train/1/images/frame_000023.png
diff --git a/tests/assets/cvat_dataset/action_detection/train/1/images/frame_000024.png b/tests/assets/cvat_dataset/action_detection/train/1/images/frame_000024.png
diff --git a/tests/assets/cvat_dataset/action_detection/train/1/images/frame_000025.png b/tests/assets/cvat_dataset/action_detection/train/1/images/frame_000025.png
diff --git a/tests/assets/cvat_dataset/action_detection/train/1/images/frame_000026.png b/tests/assets/cvat_dataset/action_detection/train/1/images/frame_000026.png
diff --git a/tests/assets/cvat_dataset/action_detection/train/1/images/frame_000027.png b/tests/assets/cvat_dataset/action_detection/train/1/images/frame_000027.png
diff --git a/tests/assets/cvat_dataset/action_detection/train/1/images/frame_000028.png b/tests/assets/cvat_dataset/action_detection/train/1/images/frame_000028.png
diff --git a/tests/assets/cvat_dataset/action_detection/train/1/images/frame_000029.png b/tests/assets/cvat_dataset/action_detection/train/1/images/frame_000029.png