The method improves the accuracy of facial recognition by selecting only the most distinctive images from a video.
The method reduces the computational load required for facial recognition, which makes the process faster and more economical.
The method can be adapted for different facial recognition applications.
Facial recognition is a technology that is increasingly being used in various contexts, such as security, access control and marketing. However, identifying individuals from video images can be a complex and computationally intensive process.
The method for selecting and managing facial images presented by PUC-Rio is an innovative solution that makes facial recognition more accurate and efficient. The method is based on rules that make it possible to select only the most distinctive facial images from a video, which reduces the computational burden required for recognition.
Facial images are extracted from each frame of the video and compared with each other to determine their similarity. A facial image is only selected for recognition if it is significantly different from the other selected images. Finally, the selected images are stored in an LRU (least recently used) queue, which discards the oldest images when the queue is full.
Patent title:
Method for selecting and managing facial images that make up the evidence in computer systems for facial recognition from video sequences
Deposit Number:
BR 11 2019 012800 0
Pontifical Catholic University of Rio de Janeiro – PUC-Rio
Rua Marquês de São Vicente, 225, Gávea
Rio de Janeiro, RJ – Brasil
Zip code: 22451-900
Postal box: 38097
Phone:
+55 (21) 3527-2155