Detect unknown and known objects using image semantics and slots with OpenSlot
Detect unknown and known objects using image semantics and slots with OpenSlot
OpenSlot: Mixed Open-set Recognition with Object-centric Learning
arXiv paper abstract https://arxiv.org/abs/2407.02386
arXiv PDF paper https://arxiv.org/pdf/2407.02386
… open-set recognition (OSR) … assume … image contains only one … label, and the unknown test set (negative) has a disjoint label space from the known test set (positive) …
… introduces the mixed OSR problem, where test images contain multiple class semantics, with known and unknown classes co-occurring in negatives
… propose the OpenSlot framework, built upon object-centric learning. OpenSlot utilizes slot features to represent diverse class semantics and produce class predictions.
… anti-noise-slot (ANS) … mitigate the … noise (invalid and background) slots during … training, … addressing … misalignment between … predictions and the ground truth.
… OpenSlot … exceeds existing OSR studies in detecting super-label shifts across single & multi-label mixed OSR tasks … achieves state-of-the-art performance on … benchmarks.
Remarkably, … method can localize class objects without using bounding boxes during training …
Stay up to date. Subscribe to my posts https://morrislee1234.wixsite.com/website/contact
Web site with my other posts by category https://morrislee1234.wixsite.com/website
LinkedIn https://www.linkedin.com/in/morris-lee-47877b7b
