official stable release

Author: nwoyecid

ivtmetrics/__init__.py

@@ -2,7 +2,7 @@
 # -*- coding: utf-8 -*-
 
 __library__ = "ivtmetrics"
-__version__ = "0.0.9"
+__version__ = "0.1.0"
 __author__  = 'Chinedu Nwoye'
 __supervisor__  = 'Nicolas Padoy'
 __credits__ = 'CAMMA, ICube Lab, University of Strasbourg, France'

ivtmetrics/recognition.py

@@ -110,9 +110,9 @@ def compute_AP(self, component="ivt", ignore_null=False):
             targets  = self.extract(self.targets, component)
             predicts = self.extract(self.predictions, component)
         else:
-            sys.exit(f"Function filtering {component} not yet supported!")
+            sys.exit("Function filtering {} not yet supported!".format(component))
         with warnings.catch_warnings():
-            warnings.filterwarnings(action='ignore', message='[info] triplet classes not represented in this test sample will be reported as nan values.')            
+            warnings.filterwarnings(action='ignore') #, message='[info] triplet classes not represented in this test sample will be reported as nan values.')            
             classwise = average_precision_score(targets, predicts, average=None)
             if (ignore_null and component=="ivt"): classwise = classwise[:-6]
             mean      = np.nanmean(classwise)
@@ -143,9 +143,9 @@ def compute_global_AP(self, component="ivt", ignore_null=False):
             targets  = self.extract(targets, component)
             predicts = self.extract(predicts, component)
         else:
-            sys.exit(f"Function filtering {component} not yet supported!")            
+            sys.exit("Function filtering {} not yet supported!".format(component))            
         with warnings.catch_warnings():
-            warnings.filterwarnings(action='ignore', message='[info] triplet classes not represented in this test sample will be reported as nan values.')            
+            warnings.filterwarnings(action='ignore') #, message='[info] triplet classes not represented in this test sample will be reported as nan values.')            
             classwise = average_precision_score(targets, predicts, average=None)
             if (ignore_null and component=="ivt"): classwise = classwise[:-6]
             mean      = np.nanmean(classwise)
@@ -178,7 +178,7 @@ def compute_video_AP(self, component="ivt", ignore_null=False):
                     targets  = self.extract(targets, component)
                     predicts = self.extract(predicts, component)
                 else:
-                    sys.exit(f"Function filtering {component} not yet supported!")                        
+                    sys.exit("Function filtering {} not yet supported!".format(component))                        
                 classwise = average_precision_score(targets, predicts, average=None)
                 video_log.append( classwise.reshape([1,-1]) )
             video_log = np.concatenate(video_log, axis=0)         
@@ -214,7 +214,7 @@ def topK(self, k=5, component="ivt"):
             targets  = self.extract(targets, component)
             predicts = self.extract(predicts, component)
         else:
-            sys.exit(f"Function filtering {component} not supported yet!")
+            sys.exit("Function filtering {} not supported yet!".format(component))
         correct = 0.0
         total   = 0
         for gt, pd in zip(targets, predicts):
@@ -248,7 +248,7 @@ def topClass(self, k=10, component="ivt"):
             targets  = self.extract(targets, component)
             predicts = self.extract(predicts, component)
         else:
-            sys.exit(f"Function filtering {component} not supported yet!")            
+            sys.exit("Function filtering {} not supported yet!".format(component))            
         classwise = average_precision_score(targets, predicts, average=None)
         pd_idx    = (-classwise).argsort()[:k]
         output    = {x:classwise[x] for x in pd_idx}

setup.py

@@ -6,20 +6,20 @@
 
 setup(
     name='ivtmetrics',
-    version='0.0.9',    
+    version='0.1.0',    
     packages=['ivtmetrics'],
     author='Chinedu Nwoye',
     author_email='nwoye.chinedu@gmail.com',    
     description='A Python evaluation metrics package for action triplet recognition',
-    keywords = ['triplet', 'average precision', 'AP'], 
+    keywords = ['triplet', 'average precision', 'AP', 'mAP'], 
     long_description = long_description,
     long_description_content_type ='text/x-rst',
     url='https://github.com/CAMMA-public/ivtmetrics',
-    download_url = 'https://github.com/CAMMA-public/ivtmetrics/archive/refs/tags/v0.0.9.tar.gz',    # I explain this later on
+    download_url = 'https://github.com/CAMMA-public/ivtmetrics/archive/refs/tags/v0.1.0.tar.gz', 
     include_package_data=True,
     license='BSD 2-clause', # Chose a license from here: https://help.github.com/articles/licensing-a-repository
     install_requires=['scikit-learn',
-                      'numpy',
+                      'numpy>=1.21',
                       ],
 
     classifiers=[