AI Paper Atlas

Research map / Detection and segmentation / Weakly supervised detection and segmentation

Diff3DETR: Agent-based Diffusion Model for Semi-supervised 3D Object Detection

Jiacheng Deng, Jiahao Lu, Tianzhu Zhang

ECCV 2024 · Detection and segmentation › Weakly supervised detection and segmentation · approach: Pseudo-labeling for semi-supervised detection · 14 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add agent-based object query generator (ablation)Diff3DETR baseline · ScanNet 5% labeledmAP@0.25: 42.2 -> 43.1Improved
add agent adaptor to the agent-based query generator (ablation)Diff3DETR · ScanNet 5% labeledmAP@0.25: 43.1 -> 43.5Improved
add box-aware denoising module (ablation)Diff3DETR · ScanNet 5% labeledmAP@0.25: 43.5 -> 44.3Improved
add 3D vertex relative position encoding (ablation)Diff3DETR · ScanNet 5% labeledmAP@0.25: 44.3 -> 44.6Improved
add DDIM iterative denoising (ablation)Diff3DETR · ScanNet 5% labeledmAP@0.25: 44.6 -> 45.1Improved
increase decoder layers from 3 to 9 (DDIM=2) (ablation)Diff3DETR · ScanNet 5% labeledmAP@0.25: 45.1 -> 46.3Improved
increase DDIM steps from 2 to 4 (DL=3) (ablation)Diff3DETR · ScanNet 5% labeledmAP@0.5: 29.2 -> 28.8No clear effect
increase decoder layers and DDIM steps beyond 3 and 2 (ablation)Diff3DETR · ScanNettraining and inference speedHurt
use agent grid resolution 16x16x4 rather than other resolutions (ablation)Diff3DETR · ScanNetmAPImproved
replace Diffusion-SS3D with Diff3DETRDiffusion-SS3D · ScanNet 5% labeledmAP@0.25: 43.5 -> 45.1Improved
replace NESIE with Diff3DETR at 20% labelsNESIE · SUN RGB-D 20% labeledmAP@0.25: 53.4 -> 53.2No clear effect

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy