Merge masterinto develop (#261)

* Synchronize master and develop workflows (#236)

* Synchronize master and develop workflows

* comment

* Added OpenDR citation (#238)

* Added OpenDR citation

* Update README.md

* Fixes bibtex name (#241)

Fixes citation name

* Fix clang (#250)

* Integration of heart anomaly detection self-attention neural bag of features  (#246)

* added sanbof models

* added attention models to ci test

Co-authored-by: ad-daniel <44834743+ad-daniel@users.noreply.github.com>

* Make `test release` docker target the specific branch when the label is run manually (#252)

* Fix

* Better approach

* Fix

* Update CODEOWNERS (#253)

* Update face-recognition.md (#257)

* skeleton-based HAR demo bug fixed (#260)

* infer updated

* edge case fix (no pose detected)

* pep fixes

* Fix sources after cppcheck upgrade

* More

Co-authored-by: Negar <negar.heidari@eng.au.dk>
Co-authored-by: Nikolaos Passalis <passalis@users.noreply.github.com>
Co-authored-by: ad-daniel <daniel.dias@epfl.ch>

* Update seq2seq_nms_learner.py

* Update seq2seq_nms_learner.py

* add init

Co-authored-by: Nikolaos Passalis <passalis@users.noreply.github.com>
Co-authored-by: Kateryna Chumachenko <katerynaCh@users.noreply.github.com>
Co-authored-by: Stefania Pedrazzi <stefaniapedrazzi@users.noreply.github.com>
Co-authored-by: Pavlos Tosidis <35866477+Pavlos-Tosidis@users.noreply.github.com>
Co-authored-by: Negar Heidari <36771997+negarhdr@users.noreply.github.com>
Co-authored-by: Negar <negar.heidari@eng.au.dk>
Co-authored-by: charsyme <63857415+charsyme@users.noreply.github.com>
Co-authored-by: charsyme <charsyme@csd.auth.gr>

.github/workflows/tests_suite_develop.yml

@@ -68,6 +68,7 @@ jobs:
           - perception/object_tracking_2d
           - perception/object_detection_3d
           - perception/pose_estimation
+          - perception/fall_detection
           - perception/speech_recognition
           - perception/skeleton_based_action_recognition
           - perception/semantic_segmentation
@@ -150,9 +151,6 @@ jobs:
       with:
         submodules: true
         ref: develop
-    - name: Build image
-      run: |
-        docker build --tag opendr/opendr-toolkit:cpu_test --file Dockerfile .
     - name: Get branch name
       id: branch-name
       uses: tj-actions/branch-names@v5.1
@@ -182,6 +180,7 @@ jobs:
           - perception/multimodal_human_centric
           - perception/object_tracking_2d
           - perception/pose_estimation
+          - perception/fall_detection
           - perception/speech_recognition
           - perception/skeleton_based_action_recognition
           - perception/semantic_segmentation
@@ -253,6 +252,7 @@ jobs:
           - perception/multimodal_human_centric
           - perception/object_tracking_2d
           - perception/pose_estimation
+          - perception/fall_detection
           - perception/speech_recognition
           - perception/skeleton_based_action_recognition
           - perception/semantic_segmentation
@@ -330,6 +330,7 @@ jobs:
           - perception/multimodal_human_centric
           - perception/object_tracking_2d
           - perception/pose_estimation
+          - perception/fall_detection
           - perception/speech_recognition
           - perception/skeleton_based_action_recognition
           - perception/semantic_segmentation
@@ -372,7 +373,7 @@ jobs:
       with:
         name: docker-artifact
   delete-wheel-artifacts:
-    needs: [build-wheel, test-wheel]
+    needs: [build-wheel, test-wheel, test-wheel-separate]
     if: ${{ always() }}
     strategy:
       matrix:

Dockerfile-cuda

@@ -6,6 +6,8 @@ ARG branch
 ENV APT_KEY_DONT_WARN_ON_DANGEROUS_USAGE=1
 RUN apt-key adv --fetch-keys https://developer.download.nvidia.com/compute/cuda/repos/ubuntu$(cat /etc/os-release | grep VERSION_ID | awk '{print substr($0,13,5)}' | awk -F'.' '{print $1$2}')/x86_64/3bf863cc.pub
 
+ARG branch
+
 # Install dependencies
 RUN apt-get update && \
     apt-get --yes install git sudo apt-utils

docs/reference/face-recognition.md

@@ -303,6 +303,7 @@ recognizer.save('./temp/saved_models')
 * **Inference example - backbone_only mode**
 ```python
 from opendr.perception.face_recognition import FaceRecognitionLearner
+from opendr.engine.data import Image
 recognizer = FaceRecognitionLearner(backbone='ir_50', mode='backbone_only', device='cuda')
 recognizer.load('./temp/saved_models')
 recognizer.fit_reference(path='./data/imgs', save_path='./temp/demo')

projects/perception/skeleton_based_action_recognition/demos/demo.py

@@ -205,10 +205,11 @@ def draw_preds(frame, preds: Dict):
             if counter > 0:
                 skeleton_seq = pose2numpy(args, counter, poses_list)
 
-            prediction = action_classifier.infer(skeleton_seq)
-            category_labels = preds2label(prediction.confidence)
-            print(category_labels)
-            draw_preds(img, category_labels)
+                prediction = action_classifier.infer(skeleton_seq)
+                category_labels = preds2label(prediction.confidence)
+                print(category_labels)
+                draw_preds(img, category_labels)
+
             # Calculate a running average on FPS
             end_time = time.perf_counter()
             fps = 1.0 / (end_time - start_time)

src/c_api/face_recognition.cpp

@@ -289,20 +289,20 @@ void build_database_face_recognition(const char *database_folder, const char *ou
   // Write number of persons
   int n = person_names.size();
 
-  fout.write((char *)(&n), sizeof(int));
+  fout.write(static_cast<char *>(&n), sizeof(int));
   for (int i = 0; i < n; i++) {
     // Write the name of the person (along with its size)
     int name_length = person_names[i].size() + 1;
-    fout.write((char *)(&name_length), sizeof(int));
+    fout.write(static_cast<char *>(&name_length), sizeof(int));
     fout.write(person_names[i].c_str(), name_length);
   }
 
   cv::Size s = database_out.size();
 
-  fout.write((char *)(&s.height), sizeof(int));
-  fout.write((char *)(&s.width), sizeof(int));
-  fout.write((char *)database_out.data, sizeof(float) * s.height * s.width);
-  fout.write((char *)(&database_ids[0]), sizeof(int) * s.height);
+  fout.write(static_cast<char *>(&s.height), sizeof(int));
+  fout.write(static_cast<char *>(&s.width), sizeof(int));
+  fout.write(static_cast<char *>(database_out.data), sizeof(float) * s.height * s.width);
+  fout.write(static_cast<char *>(&database_ids[0]), sizeof(int) * s.height);
   fout.flush();
   fout.close();
 }
@@ -318,14 +318,14 @@ void load_database_face_recognition(const char *database_path, face_recognition_
     return;
   }
   int n;
-  fin.read((char *)(&n), sizeof(int));
+  fin.read(static_cast<char *>(&n), sizeof(int));
   char **person_names = new char *[n];
 
   for (int i = 0; i < n; i++) {
     person_names[i] = new char[512];
     // Read person name
     int name_length;
-    fin.read((char *)(&name_length), sizeof(int));
+    fin.read(static_cast<char *>(&name_length), sizeof(int));
     if (name_length > 512) {
       std::cerr << "Person name exceeds max number of characters (512)" << std::endl;
       return;
@@ -334,13 +334,13 @@ void load_database_face_recognition(const char *database_path, face_recognition_
   }
 
   int height, width;
-  fin.read((char *)(&height), sizeof(int));
-  fin.read((char *)(&width), sizeof(int));
+  fin.read(static_cast<char *>(&height), sizeof(int));
+  fin.read(static_cast<char *>(&width), sizeof(int));
 
   float *database_buff = new float[height * width];
   int *features_ids = new int[height];
-  fin.read((char *)(database_buff), sizeof(float) * height * width);
-  fin.read((char *)(features_ids), sizeof(int) * height);
+  fin.read(static_cast<char *>(database_buff), sizeof(float) * height * width);
+  fin.read(static_cast<char *>(features_ids), sizeof(int) * height);
 
   fin.close();
 

src/opendr/perception/skeleton_based_action_recognition/progressive_spatio_temporal_gcn_learner.py

@@ -566,10 +566,9 @@ def infer(self, skeletonseq_batch):
 
         m = nn.Softmax(dim=0)
         softmax_predictions = m(output.data[0])
-        class_confidence = float(torch.max(softmax_predictions))
         class_ind = int(torch.argmax(softmax_predictions))
         class_description = self.classes_dict[class_ind]
-        category = Category(prediction=class_ind, confidence=class_confidence, description=class_description)
+        category = Category(prediction=class_ind, confidence=softmax_predictions, description=class_description)
 
         return category
 

src/opendr/perception/skeleton_based_action_recognition/spatio_temporal_gcn_learner.py

@@ -530,10 +530,9 @@ def infer(self, SkeletonSeq_batch):
 
         m = nn.Softmax(dim=0)
         softmax_predictions = m(output.data[0])
-        class_confidence = float(torch.max(softmax_predictions))
         class_ind = int(torch.argmax(softmax_predictions))
         class_description = self.classes_dict[class_ind]
-        category = Category(prediction=class_ind, confidence=class_confidence, description=class_description)
+        category = Category(prediction=class_ind, confidence=softmax_predictions, description=class_description)
 
         return category