Implementation of OFA-DOD

[xiaofeng94]

Dear authors,

Thanks for your cool job of "Described Object Detection: Liberating Object Detection with Flexible Expressions". I'm impressed by the D^3 dataset and the proposed method OFA-DOD.
However, I only found instructions for the dataset. Did you release the code for OFA-DOD somewhere?

Best.

[Charles-Xie]

Hi, thanks for your interest!
We have not released the code for the OFA-DOD baseline.
However, we have just released the code on some other baselines (SPHINX, UNINEXT, G-DINO), and a page for the leaderboard on D3: https://github.com/shikras/d-cube/tree/main/eval_sota. Please have a look if you are interested.
We will continue to work to release the code for more baselines.

If it may interest you, we have also evaluated the MLLM, SPHINX, too, which is the first MLLM benchmarked for DOD.
We are also impressed by your work, generating enhanced negatives, and we have added it in the page too.

Best,
Chi