Computer Vision News - January 2021

327 Spotlight News Top 5 Computer Vision Trends That Will Rule 2021 This is not an article for the most technical and experienced among us. From Embedded Vision to Virtual Reality, from Edge Computing to Augmented Reality and Computer Vision As A Service: many are the advances in the field of computer vision, mainly by joining forces with different technologies to boost trends that will dominate the year of 2021. If you are new in the computer vision arena, you should definitely read this simple article. The good news is that both exposure and investment in computer vision keep growing. Read More Forgetting in Deep Learning Apparently, a model can drastically lose its generalization ability on a task after being trained on a new task. This phenomenon is called catastrophic forgetting . This usually means that a new task will likely override the weights that have been learned in the past and thus degrade the model performance obtained for the past tasks. When this problem is not fixed, a single neural network will not be able to adapt itself in cases of continuous learning scenario, because it forgets the existing information/knowledge when it learns new things. The authors do not suggest a fit-for-all solution, but find some interesting results related to data augmentation and learning rates. Read More Is the Biopharma Industry Right to Be Skeptical about AI? Here’s one more article with a counter-intuitive thesis and a somehow provocative title. It seems that on one hand, some experts claim that AI could revolutionize biopharma and help us to discover new treatment options, with Deloitte predicting that the AI/biopharma industry will be worth $3.88 billion by 2025. The opposite view warns biopharma to be skeptical because there’s so much noise and hype in this game. We are glad to read that among the good news that image processing software has fulfilled his promises of contributing to healthcare by partnering with humans to carry out safety checks and manage human oversights. Read More