filter_tiny_segments

README.md

@@ -1,2 +1,71 @@
-# filter-tiny-segments
+# Filter Tiny Segments
 
+*author: Chen Zhang*
+
+
+<br />
+
+
+
+## Description
+
+This operator can filter tiny detected segments with format of list of `[start_second_1, start_second_2, end_second_1, end_second_2]`
+<br />
+
+
+## Code Example
+
+
+```python
+import towhee
+towhee.dc['pred']([[[0, 0, 100, 100], [0, 0, 10, 10], [0, 0, 60, 10]]]) \
+      .video_copy_detection.filter_tiny_segments['pred', 'filtered_pred'](filter_s_thresh=20) \
+      .show()
+```
+
+![](result.png)
+
+
+
+## Factory Constructor
+
+Create the operator via the following factory method
+
+***filter_tiny_segments(filter_s_thresh, segment_len_rate)***
+
+
+**Parameters:**
+
+​   ***filter_s_thresh:*** *float*
+
+​   Use a thresh to filter detected box which is smaller than it.
+
+​   ***segment_len_rate:*** *float*
+
+​   Filter expect longer then segment_len_rate * video length. Only useful for filter expect near video length segments.
+
+<br />
+
+
+
+## Interface
+
+A Temporal Network operator takes two numpy.ndarray(shape(N,D) N: number of features. D: dimension of features) and get the duplicated ranges and scores.
+
+
+**Parameters:**
+
+​	***pred_value:*** *List*
+
+​  List of predicted segment second infos of a video pair
+
+​  ***sim_hw:*** *Tuple*
+
+​  Similarity matrix height and weight of a video pair. If sample rate is 1s, sim_hw is also the lengths of these videos.
+
+
+**Returns:**
+
+​	***res_pred_list:*** *List*
+
+​   List of filtered predicted segment second infos

__init__.py

@@ -0,0 +1,19 @@
+# Copyright 2022 Zilliz. All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+from .filter_tiny_segments import FilterTinySegments
+
+
+def filter_tiny_segments(filter_s_thresh: float, segment_len_rate: float):
+    return FilterTinySegments(filter_s_thresh, segment_len_rate)

filter_tiny_segments.py

@@ -0,0 +1,32 @@
+from typing import List, Tuple, Optional
+from towhee.operator.base import Operator
+from towhee import register
+
+
+@register(output_schema=['vec'])
+class FilterTinySegments(Operator):
+    """
+    FilterTinySegments
+    """
+
+    def __init__(self, filter_s_thresh: float, segment_len_rate: float = 0.8):
+        super().__init__()
+        self.filter_s_thresh = filter_s_thresh
+        self.segment_len_rate = segment_len_rate
+
+    def __call__(self, pred_value: List[Optional[List]], sim_hw: Tuple = None) -> List:
+        res_pred_list = []
+        for time_info in pred_value:
+            filterable = True
+            if ((time_info[2] - time_info[0]) + (time_info[3] - time_info[1])) / 2.0 > self.filter_s_thresh:
+                filterable = False
+            if sim_hw is not None:
+                sim_h, sim_w = sim_hw
+                if (time_info[2] - time_info[0]) > sim_h * self.segment_len_rate or (
+                        time_info[3] - time_info[1]) > sim_w * self.segment_len_rate:
+                    filterable = False
+            if not filterable:
+                res_pred_list.append(time_info)
+        return res_pred_list
+
+