====== Human Pose Estimation ======
{{ ::tennis_gesture.png?600 |}}

<WRAP center round box 80% centeralign>
**{{tagpage>skeleton camera3D|Les pages sur les caméras 3D et la détection de squelette}}**     **[[http://translate.google.com/translate?hl=&sl=auto&tl=en&u=https%3A%2F%2Fressources.labomedia.org%2Fhuman_pose_estimation|English Version]]**
</WRAP>
{{ :media_15:img_0005.jpg?600 |}}
=====Ressources=====
====Wikipedia====
  * [[https://en.wikipedia.org/wiki/Pose_(computer_vision)|Pose (computer vision)]] @ en.wikipedia.org: In computer vision and robotics, a typical task is to identify specific objects in an image and to determine each object's position and orientation relative to some coordinate system.
  * [[https://en.wikipedia.org/wiki/3D_pose_estimation|3D pose estimation]] @ en.wikipedia.org: 3D pose estimation is a process of predicting the transformation of an object from a user-defined reference pose, given an image or a 3D scan. 
  * [[https://en.wikipedia.org/wiki/Gesture_recognition|Gesture recognition]] @ en.wikipedia.org: Gesture recognition is a topic in computer science and language technology with the goal of interpreting human gestures via mathematical algorithms.


====Les pages sur ce wiki====
{{topic>skeleton camera3D}}


=====Des collections de données, des concours=====
====Principes====
  * **[[https://opencv.org/|OpenCV.org]]** 
  * **[[https://opencv.org/opencv-ai-competition-2021/#phase1-winners-list|OpenCV AI Competition]]** Un concours est en cours en Mars 2021
  * **[[https://www.tensorflow.org/|TensorFlow.org]]**
  * Réseau de neurones: **[[https://fr.wikipedia.org/wiki/R%C3%A9seau_de_neurones_r%C3%A9currents|RNN]]** @fr.wikipedia.org
  * Convolution: **[[https://fr.wikipedia.org/wiki/R%C3%A9seau_neuronal_convolutif|CNN]]** @fr.wikipedia.org
====Des datas====
  * **[[https://en.m.wikipedia.org/wiki/List_of_datasets_for_machine-learning_research|List of datasets for machine-learning research]]** @ wikipedia.org

====Data Sets====
===COCO===
  * [[https://cocodataset.org/#home|Common Object in Context]]. COCO is a large-scale object detection, segmentation, and captioning dataset.
  * [[https://cocodataset.org/#keypoints-2019|COCO 2019 Keypoint Detection Task]] **Le concours de détection de squelette** Je n'ai pas trouvé le gagnant :?:

===MPI Human Pose===
  * [[http://human-pose.mpi-inf.mpg.de/|human-pose du Max Plant Institut]]

===Human Pose Evaluator Dataset ===
  * [[https://www.robots.ox.ac.uk/~vgg/data/pose_evaluation/|MPII Human Pose dataset is a state of the art benchmark for evaluation of articulated human pose estimation]]

====Les normes COCO et MPI====
<WRAP group>
<WRAP third column>
{{:media_14:skeleton_kinect.png?200|}}\\
kinect
</WRAP>
<WRAP third column>
{{:media_14:mpi-keypoints.png?400|}}
</WRAP>
<WRAP third column>
{{:media_14:coco-keypoints.png?400|}}
</WRAP>
</WRAP>
=====Solutions matérielles et logicielles=====
  * **[[https://towardsdatascience.com/nvidia-jetson-nano-vs-google-coral-vs-intel-ncs-a-comparison-9f950ee88f0d|NVIDIA Jetson Nano vs Google Coral vs Intel NCS: A Comparison]]** It is evident from the latency point of view, Nvidia Jetson Nano is performing better ~25 fps as compared to ~9 fps of google coral and ~4 fps of Intel NCS.

====BlobFromImage BLOB binary large object====
  * **[[https://www.pyimagesearch.com/2017/11/06/deep-learning-opencvs-blobfromimage-works/|Deep learning: How OpenCV’s blobFromImage works]]** Explications détaillées de la théorie et de l'utilisation de OpenCV **blobFromImage**.
  * [[https://techterms.com/definition/blob|blob @ techterms.com]]
  * **[[https://en.wikipedia.org/wiki/Binary_large_object|BLOB @ en.wikipedia.org]]** et **[[https://fr.wikipedia.org/wiki/Binary_large_object|BLOB @ fr.wikipedia.org]]** Le BLOB, pour binary large object, est un type de donnée permettant le stockage de données binaires (le plus souvent des fichiers de type image, son ou vidéo) dans le champ d'une table d'une base de données. 


===== Caméra de profondeur  =====
====Intel RealSense====
  * **[[cameras_de_profondeur|]]**  Quel matériel choisir ?

====Kinect====
  * **{{tagpage>kinect|Les pages sur la Kinect sur ce wiki.}}**
  * **[[kinect_dans_blender|]]** Une visualisation d'un squelette dans le Blender Game Engine.

===== Human Pose Estimation =====
====Les essais publiés sur ce wiki====
  * **[[coral_usb_accelerator|]]** La solution avec TensorFlow et des fichiers de poids optimés pour Android
  * **[[pose_estimation_avec_opencv|]]** Les fichiers de poids sont trop lourd !
  * **[[pose_estimation_avec_intel_ncs2_et_openvino|]]** Des essais et recherches pour augmenter le FPS et ne pas utiliser les CPU et GPU.

=====D'autresqs projet Open====
  * **[[https://github.com/CMU-Perceptual-Computing-Lab/openpose|OpenPose]]** de CMU Perceptual Computing Lab @GitHub: Real-time multi-person keypoint detection library for body, face, hands, and foot estimation **with only one camera**. Whole-body (Body, Foot, Face, and Hands) 2D Pose Estimation and Whole-body 3D Pose Reconstruction and Estimation. Utilise les caffemodel identique à ceux de OpenCV !!
  * **[[http://openptrack.org/overview-3d-skeleton-tracking-pose-recognition-with-opt/|3D Skeleton Tracking et Pose Recognition with OPT]]** @ openptrack.org 
    * Pas d'API python
    * [[https://github.com/OpenPTrack/open_ptrack_v2/wiki|OpenPTrack wiki]]
    * The pose recognition module comes with a set of 3 pre-recorded poses

====Solutions Propriétaires====
  * **[[https://github.com/3DiVi/nuitrack-sdk|Nuitrack™ 60 €]]** is a 3D tracking middleware developed by 3DiVi Inc. This is a solution for skeleton tracking and gesture recognition that enables capabilities of Natural User Interface (NUI) on Android, Windows, and Linux.
  * **[[skeleton_tracking_de_cubemos_logiciel_proprietaire|Cubemos 75 €]]** Détection correcte mais propriétaire et pas du tout pour l'éternité !


Reconnaissance gestuelle

{{tag>sb skeleton camera3D}}