Computer Vision News - June 2020

Summary Artificial Intelligence 30 potlight News 46 Early Bird uses 10 times less energy to train deep neural networks What is Early Bird , you may ask? It is an energy-efficient method for training deep neural networks (DNNs). Since we all know how training is getting always costlier (the number of computations needed to train a top- flight deep neural network increased 300,000 times between 2012-2018), it is key to find greener training methods, that both address environmental concerns and reduce financial barriers of AI research. This method was found by Rice and Texas A&M University researchers and presented at ICLR 2020 . Apparently, Early Bird could use 10.7 times less energy to train a DNN to the same level of accuracy or better than typical training, which makes it worth a try. Read More Intel/MobilEye Promises Self-Driving Robotaxi Service In 2022, While Others Back Off Finally, an article about autonomous driving vehicles, which lately have become much less popular than they used to be. Apparently, many actors in the automotive arena are being more cautious with promises and announcements. The exception is Amnon Shashua , founder and CEO of MobilEye , now a unit of Intel INTC , who just declared their intention to offer robotaxi service, with no safety drivers, in early 2022. Which is just after tomorrow. This happens while we see layoffs at several companies, including some of the biggest start-ups like Zoox and Cruise . 2022 is not a new horizon for MobilEye, but it is surprising that they confirm it after a difficult year for ADAS research. Shashua also predicted there will be significant consolidation among players. Read More Deep learning for OCT image analysis Millions of optical coherence tomography (OCT) images are acquired each year, providing anatomical insights to diagnose numerous medical conditions, including diabetic macular edema (DME) . This story is about researchers at Vienna University who used deep learning to segment ophthalmic OCT images in order to recognize the tear meniscus , which is the layer of liquid present as the eyelid traverses the surface of the eye. This measure enables to diagnose patients suffering from dry-eye. RSIP Vision has done during the years many projects on sophisticated ophthalmic imaging and measurement techniques using Deep Learning and AI: you can find them here. Read More