Reproduce “CL3DOD: Understanding the Stability-Plasticity Dilemma in Outdoor Class-Incremental 3D Object Detection” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Muhammad Ahmed Ullah Khan, Khalil Ahmed, Didier Stricker, Muhammad Zeshan Afzal

CL3DOD updates LiDAR detectors incrementally, adding categories without old annotations. It uses a previous-stage teacher to generate confidence- and overlap-filtered pseudo-labels for old objects, plus query-wise distillation and temporal consistency. In three-task nuScenes validation after Task 3, it reaches 55.71 current-task mAP and 55.45 cumulative mAP, versus 1.43 and 5.22 for same-detector Freeze-and-Add; these results reflect the tested split.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.