DETR- End-to-End Object Detection with Transformers - Paper Explained
Aleksa Gordić - The AI Epiphany via YouTube
Overview
Google, IBM & Meta Certificates – 40% Off
One Coursera Plus subscription covers most Professional Certificates on Coursera.
Unlock All Certificates
This course explains DETR, an end-to-end object detection pipeline that uses a transformer encoder-decoder and Hungarian bipartite matching loss to produce unique predictions without non-maximum suppression. It also reviews architecture details, ablations, visualizations, and an extension to panoptic segmentation.
Syllabus
Intro: DETR main ideas
Non-max suppression
High-level pipeline overview
Architecture in more detail
Matching loss
Hungarian loss
Results
Visualization
Ablations
Outro: Segmentation results
Taught by
Aleksa Gordić - The AI Epiphany