Publication · August 25, 2026
Research Note: Detecting Faint Exoplanets
A concise record of the 2024 BRAC University work co-authored by Ashraful Kabir Alif and Shihab Shahriar Antor.
By Shihab Shahriar Antor · Updated 2026-08-25
A 2024 BRAC University research work studied contrastive and few-shot learning for detecting faint exoplanets and stars in direct-imaging data. The authors are Ashraful Kabir Alif, S. J. Hossain, M. I. Hossain, S. S. Antor, and A. P. Roy.
The problem
Direct imaging is difficult because a bright host star can overwhelm the much weaker planetary signal. The work uses JWST-related astronomical imagery and combines representation learning with object-detection approaches.
The reported result
The repository abstract reports 92.16% accuracy with YOLOv5, alongside Dice Coefficient and Intersection over Union measurements. Accuracy should not be restated as precision or recall; those metrics answer different questions.
How Shahriar Labs cites it
This is a co-authored university research output, not sole-authored work. Shihab Shahriar Antor is credited as S. S. Antor, and Ashraful Kabir Alif is the first author. The BRAC University repository record is the canonical source.