Shijie Wu

Mar 17, 2024 | medium.com | Shijie Wu

We discuss the common problem faced by many image transformers, large memory consumption and slow convergence speed, and discuss some historic (2 years old) approaches that have made them more efficient to train and infer. Works covered in this article include DETR, Deformable-DETR, Mask2Former, and Pix2Seq. Object detection is a task that given an image, the model classifies and detects all objects in that image.

Oct 21, 2023 | mdpi.com | Shuzhen Zhang |Shijie Wu |Lun Jia |Jiongde Jin

This is an early access version, the complete PDF, HTML, and XML versions will be available soon. Open AccessArticleby *, , , and School of Mechanical and Electrical Engineering, Lanzhou University of Technology, Lanzhou 730050, China*Author to whom correspondence should be addressed.