Indoor-Outdoor Image Retrieval
Laura discusses the challenges of computing relative poses for images in different environments, emphasizing the need for specialized models for indoor and outdoor settings. She explains their approach to fixing scale issues in essential matrices, allowing the network to handle varying scales effectively. The ongoing research aims to determine whether a single network can be adapted or if distinct models are necessary for optimal performance.In this clip
From this podcast
The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence)
Dynamic Visual Localization and Segmentation with Laura Leal-Taixé -TWiML Talk #168
Related Questions