Object Detection with Facebook Detr Transformer Model

Iamranjanbatra
3 min readMar 11, 2024

DeTr model, a class of transformer model and trained on coco dataset of images. This dataset has about 118k annotated images.

DeTr model has three key components.
1. CNN Backbone (Resnet)
CNN Backbone basically stands for convolutional neural network backbone, which is a foundational block for Convolutional neural network.
This is used for feature extraction for various…

--

--