Title: P3: preparing for interview and reading paper [2024-02-28 Wed]
detection networks. It uses predefined anchor boxes and their
pyramides. There is a sliding window, a box-regression layer
(reg) and a box-classification layer (cls).
Anchor-free object detection methods is CenterNet, FCOS
(Fully Convolutional One-Stage Object Detection) and
DETR (DEtection TRansformers)
😶 #dailyreport #cv #objectdetection #fsl #deeplearning
Title: P2: preparing for interview and reading paper [2024-02-28 Wed]
- Learn-to-Parameterize - param eterizing the base learner or
some subparts of base learner for a novel task so that it can
address this task specifically. meta learner generate weights
for base learner.
- Learn-to-Adjust
- Learn-to-Remember
Also this article have good overview of all ML tasks.
Region Proposal Network (RPN) is a backbone of first object #dailyreport #cv #objectdetection #fsl #deeplearning
Title: P0: preparing for interview and reading paper [2024-02-28 Wed]
Few shot learning (FSL):
- 2023 A Survey on Machine Learning from Few Samples
CV Object detecttion:
- 2016 Faster R-CNN: Towards Real-Time Object
Detection with Region Proposal Networks
- 2018 Mask R-CNN
- 2015 YOLO
Most solutions for FSL in non-deep period before 2015
was generative based, but then discriminative.
Discriminative approaches is: #dailyreport #cv #objectdetection #fsl #deeplearning
anoncheg
Mathrubhumi English
Daniel Cerqueira
Guillaume Noblet
wiwsport
Pascal