|
|
Face Detection in Poor Quality Videos
Koval, Michal ; Orság, Filip (referee) ; Goldmann, Tomáš (advisor)
This bachelor thesis deals with face detection in low quality videos, while mainly focusing on occluded faces. It describes elementary priciples of machine learning algorithms and their methods, which are often used in the field of computer vision. Out of them are more closely described convolutional neural networks and their state of the art models focused on face detection. Out of those, convolutional neural networks and state of the art models for face detection are more closely described. For the practical part face detection models inspired by state of the art model RetinaFace were implemented and trained. The best performing model achieves 85.5% average precision on WIDER Face HARD testing dataset and 90.9% on dataset focused on occluded faces. Part of this thesis is also a program with graphical user interfaces which provides tools to use developed models on videos and pictures.
|
|
|
People recognition using facial images
Lindovský, Michal ; Burget, Radim (referee) ; Rajnoha, Martin (advisor)
This bachelor thesis focuses on the person recognition between several millions of people in a few seconds. As a part of my thesis is comparison of two programs which are used for recognizing faces - OpenFace and Face Recognition. Computing times of localization and face encoding are compared. The accuracy of recognition in various tests is compared as well, such as blurred image, brightness changes, age of person or usage of sunglasses. Created web application is made for recognizing people in different databases. Is possible to add or remove databases of people in the application. The application allows to subsume people into database by gender automatically or manually. Face recognition can be speeded up by using multiple processor cores.
|
|
|
Head Pose Estimation in an Image by a Neural Network
Rybnikár, Lukáš ; Goldmann, Tomáš (referee) ; Orság, Filip (advisor)
Artificial neural networks are not a novelty, but their recent rise in popularity is noticeable as well as their gain of attention from the masses. This bachelor thesis focuses on the head pose estimation in an image using the convolution neural networks. The fields of use of neural networks are vast and during last years strong enough hardware has been developed to allow us to train these networks under commonly accessible conditions. In theoretical part there are neural networks introduced with an explanation of what they are, how they work, how they are divided followed by a detailed description of convolutional neural networks. In the practical part the necessary tools used for development needed to perform experiments, such as determining appropriate configuration for neural network and optimization to get the best results possible, are described.
|
|
|
User Interface with Mobile Device and Camera
Rajca, Tomáš ; Zahrádka, Jiří (referee) ; Zemčík, Pavel (advisor)
The thesis describes the development of the user interface that uses the camera. By winking the left or right eye it is possible to emulate to two keys on the keyboard. For arranging the video it uses the library OpenCV. User interface is created by means of library Qt. This disertation includes an explanation of the basic propertis of the video, an explanation of principles of the video processing, overview of software applications, the software solution it selfs and implemenation of the application. In conclusion of this dissertaton resulsts are evaulated and potential further development is described.
|
|
|
Acceleration of a Neural Network for Face Detection in Low Light Conditions
Orava, Vojtěch ; Malinka, Kamil (referee) ; Goldmann, Tomáš (advisor)
The goal of this thesis is to build neural network for face detection in low light conditions, accelerate this network and compare it with other existing networks. Detection problem is solved with convolution neural network (CNN), which is trained on WIDER FACE and DARK FACE datasets. This CNN is accelerated by device Intel Neural Compute Stick 2. This work also summarise existing approaches in face detection (classic and neural networks based) and compares this approaches to the new ones. New detectors are based on TensorFlow Object Detection API. The best new model has average precision 47.1 % on custom validation dataset (detector YOLOv7-face has 42.8 % average precision). Speed of detection and influence of acceleration and quantization were also measured. With quantization some models could speed-up 3 times. Within this work, a GUI application for models tests was also developed. It can detect faces with newly created models and with some existing approaches.
|
|
|
Detecting a Partially Obscured Face in Image Data
Kedra, David ; Sakin, Martin (referee) ; Goldmann, Tomáš (advisor)
This bachelor thesis deals with the analysis of problems related to detecting partially occluded faces in camera systems and discusses current machine learning detectors. The aim was to find solutions useful for detecting hardly visible faces. For this reason, artificial occlusions were generated into datasets for training the YOLOv7, YOLOv7-tiny and RetinaNet models. A computer application that uses these detectors is presented. The models are compared with existing solutions in terms of quality and speed. The trained models perform better on most test datasets. YOLOv7 is the most accurate on the modified WIDER FACE and UFDD datasets with average precisions of 86 % and 89 % at a minimum IoU of 50 %. On the third dataset with face masks, the existing detector RetinaFace outperformes the others. According to the speed/quality ratio, YOLOv7-tiny is the most effective.
|
|
|
Face Image Frontalization Application
Tichý, Filip ; Malinka, Kamil (referee) ; Goldmann, Tomáš (advisor)
This work focuses on implementing an application for face frontalization using the CFR-GAN project and rotating the 3D face model followed by rendering. The aim of this work is to evaluate the impact of the application on face recognition accuracy based on the Fidentis dataset. The results are presented in the form of box plots, which depict the Euclidean distances between the generated frontalized images and the real images. It was found that when frontalizing using the rotation of a 3D model from high angles of rotation, the success of facial recognition process increases. Conversely, when frontalizing using the Complete Face Recovery GAN projekt, the recognition success signiĄcantly decreases. The VGG Face algorithm was used for comparing the images. The entire application is implemented in Python using commonly available libraries.
|
|
|
A computer vision system for emotion recognition
Wójcik, Jan ; Bilík, Šimon (referee) ; Janáková, Ilona (advisor)
The term paper deals with the design of an emotion recognition system, which will be used as a communication tool for people with autism spectrum disorder. Camera data will be used for emotion recognition, so it will be a computer vision application. The work deals with areas such as face detection, extraction of relevant features, finding a suitable dataset or designing a classifier.
|
|
|
Automated Human Recognition From Image Data
Dobiš, Lukáš
This paper describes an approach for automated human recognition by using convolutional neural networks (CNN) to perform facial analysis of persons face from image data. The predicted biometric indicators are following: age, gender, facial landmarks and facial expression. Network architectures with pretrained weights for each task are described. Script of interconnected CNN is explained and its results support further proposed expansion plans for live video inference.
|
|
|
Face Detection in Poor Quality Videos
Koval, Michal ; Orság, Filip (referee) ; Goldmann, Tomáš (advisor)
This bachelor thesis deals with face detection in low quality videos, while mainly focusing on occluded faces. It describes elementary priciples of machine learning algorithms and their methods, which are often used in the field of computer vision. Out of them are more closely described convolutional neural networks and their state of the art models focused on face detection. Out of those, convolutional neural networks and state of the art models for face detection are more closely described. For the practical part face detection models inspired by state of the art model RetinaFace were implemented and trained. The best performing model achieves 85.5% average precision on WIDER Face HARD testing dataset and 90.9% on dataset focused on occluded faces. Part of this thesis is also a program with graphical user interfaces which provides tools to use developed models on videos and pictures.
|
guest ::
