AI Paper Atlas

Research map / Point clouds / Segmentation and detection

3D Instances as 1D Kernels

Yizheng Wu, Min Shi, Shuaiyuan Du, Hao Lu, Zhiguo Cao, Weicai Zhong

ECCV 2022 · Point clouds › Segmentation and detection · approach: Scene completion and instance segmentation · 36 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
use LN-NMS instead of random candidates (ablation)DKNet · ScanNetV2 valAP: 48.0 -> 50.8Improved
use LN-NMS instead of NMS (ablation)DKNet · ScanNetV2 valAP@50: 64.6 -> 66.7Improved
add aggregation loss during training only (ablation)DKNet · ScanNetV2 valAP@50: 62.6 -> 64.5Improved
aggregate duplicate candidates at inference too (ablation)DKNet · ScanNetV2 valAP@50: 64.5 -> 66.7Improved
aggregate candidate features instead of eliminating duplicates (ablation)DKNet · ScanNetV2 valAP@50: 65.2 -> 66.7Improved
use features from all points of instance instead of aggregated candidates (ablation)DKNet · ScanNetV2 valAP@50: 66.7 -> 67.0No clear effect
add coordinate information to kernel (ablation)DKNet · ScanNetV2 valmAP: 49.5 -> 50.8Improved
add shape information (foreground/background pooling) to kernel (ablation)DKNet · ScanNetV2 valmAP: 49.3 -> 50.8Improved
vary kernel shape/size from [8,1] to [16,16,1] (ablation)DKNet · ScanNetV2 valmAP: 50.6 -> 50.9No clear effect
add candidate aggregation (ablation)DKNet · ScanNetV2 val detectionAP@50: 57.9 -> 59.0Improved
use CUDA LN-NMS instead of pythonDKNet · inference timems: 187 ms -> 79 msImproved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 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