SeaBird: Segmentation in Bird's View with Dice Loss Improves Monocular 3D Detection of Large Objects

Liu Ren, Xiaoming Liu, Abhinav Kumar, Xinyu Huang, Yuliang Guo

Code that ran

We have not lifted any functions out of this paper's repositories yet, so there is nothing we have run. If it links a repository, it is listed below.

Repositories linked to this paper

abhi1kumar/seabird — marked official in Papers with Code; named in the paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Improve KITTI-360 Val SoTA. (b) Improve nuScenes Val SoTA. (c) Theory Advancement. Figure 1. Teaser (a) SoTA frontal detectors struggle with large objects (low APLrg) even on a nearly balanced KITTI-360 dataset (Skewness in Fig. 7). Our proposed SeaBird achieves significant Mono3D improvements, particularly for large objects. (b) SeaBird also improves two SoTA BEV detectors, BEVerse-S [116] and HoP [121] on the nuScenes dataset, particularly for large objects. (c) Plot of convergence variance Var(ϵ) of dice and regression losses with the noise σ in depth prediction. The y-axis denotes the deviation from the optimal weight, so the lower the better. SeaBird leverages dice loss, which we prove is more noise-robust than regression losses for large objects.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2403.20318")
get_code_for_paper("2403.20318")
have("2403.20318")

Connect an agent — have() is free.