Face recognition technology faces challenges such as extreme illumination, occlusion, and varied expressions, with recent advancements addressing some issues through deep learning techniques, while challenges like face alignment and image super-resolution remain. Adrian Bulat discussed at the Perceive 2020 conference how face alignment involves accurately localizing facial features using nodal points, with techniques like 3D imaging and convolutional neural networks being explored, though limitations persist due to real-life constraints. Face super-resolution has applications in surveillance and upgrading low-resolution content, employing methods like cascading frameworks and neural networks to enhance image quality, yet struggles with blurry outputs when high-quality images are unavailable. The field of face recognition is evolving, with approaches such as "vanilla classification" and innovations like ARCFACE improving accuracy through cosine loss and angle calculations between features and weights. Pre-trained networks like FAN integrate with face recognition networks to enhance performance, though high GPU usage poses a challenge, which is mitigated by "binarization" for increased efficiency and speed.