Which of these facial recognition research should one invest his/her time and resources on? A facial recognition algorithm that is able to learn and evolve with any form of physiological adjustments/changes in the targeted object patterns or a facial recognition algorithm that is completely static and not able to adapt to the dynamism of facial adjustment that comes with either aging or accidental obstructions in patterns that might differ from the original image which the algorithm recognizes.
Dear Absalom. Thank you for your question. Face recognition can be static as it is the case for images on ID or dynamic as it is the case in video images. In the first case, the system compares the face with an identical image stored in the system. In the second case, the algorithm is designed to compare information about the location and appearance of specific facial landmarks (such as bony prominences). Through a multi-layer design, the system identifies the faces using appropriate classification algorithms and post-processes the results using model-based schemes and logistic feedback. In this case, the system has several stored facial images to compare against. The use of Convolutional Neural Networks in this area has shown acceptable sensitivity and specificity, and several modifications have been added to enhance the effectiveness of the system algorithm. I enclose a review article. I hope this helps.
A facial recognition algorithm that is able to evolve sounds good. Great work. Keep it up.
- Badges
- Science topic
- Similar topics
- Computer Science
- Images