AI Paper Atlas

Research map / Point clouds / Point cloud classification

To Supervise or Not to Supervise: Understanding and Addressing the Key Challenges of Point Cloud Transfer Learning

Souhail Hadgi, Lei Li, Maks Ovsjanikov

ECCV 2024 · Point clouds › Point cloud classification · approach: Self-supervised point cloud pretraining · 5 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 normal-prediction regularization on the first two layers to supervised pre-trainingDGCNN supervised pre-training · ModelNet40accuracy: 92.78 -> 93.34Improved
add normal-prediction regularization to supervised pre-trainingDGCNN supervised pre-training · ScanObjectNNaccuracy: 84.94 -> 85.91Improved
add normal-prediction regularization to supervised pre-trainingDGCNN supervised pre-training · S3DISIoU: 49.07 -> 49.88Improved
add normal-prediction regularization to contrastive pre-trainingDGCNN contrastive pre-training · ModelNet40accuracy: 92.78 -> 93.18Improved
add normal-prediction regularization to contrastive pre-trainingDGCNN contrastive pre-training · ScanObjectNNaccuracy: 85.05 -> 85.39Improved
regularize early layers instead of only the output layer (ablation)geometric regularization · downstream transferdownstream performanceImproved
use supervised instead of contrastive pre-training under linear probingpoint cloud pre-training · ModelNet40, ScanObjectNNaccuracyImproved
use contrastive instead of supervised pre-training under full fine-tuningDGCNN, MinkowskiNet, PCT · fine-tuning on ModelNet40, ScanObjectNNaccuracyImproved
supervised pre-training before S3DIS fine-tuningPointNet · S3DISIoU: 46.21 -> 37.46Hurt
contrastive pre-training before S3DIS fine-tuningMinkowskiNet · S3DISIoU: 66.89 -> 60.97Hurt
contrastive pre-training before S3DIS fine-tuningPointMLP · S3DISIoU: 56.59 -> 58.00Improved

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