AI Paper Atlas

Research map / Detection and segmentation / Open-set and few-shot detection

Object Discovery and Representation Networks

Olivier J. Hénaff, Skanda Koppula, Evan Shelhamer, Daniel Zoran, Andrew Jaegle, Andrew Zisserman, João Carreira, Relja Arandjelović

ECCV 2022 · Detection and segmentation › Open-set and few-shot detection · approach: Unsupervised object and part discovery · 65 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 K=16 instead of K=8 segments in k-means (ablation)Odin · COCO object discovery (ResNet-50)ABO instance: 38.3 -> 43.0Improved
use K=32 instead of K=16 segments (ablation)Odin · COCO object discovery (ResNet-50)ABO instance: 43.0 -> 43.1No clear effect
use continuous discovery-network updates (1e-3) instead of every 100 epochs (ablation)Odin · COCO Mask-RCNN 1xAP bbox: 42.6 -> 42.9Mixed
continuous updates (1e-3) versus discrete updates for object discovery (ablation)Odin · COCO object discovery (ResNet-50)ABO instance: 43.0 -> 41.3Mixed
replace hand-crafted segmentation (DetCon) with learned object discoveryDetCon_B · COCO object discovery (ResNet-50)ABO instance: 34.1 -> 41.5Improved
replace hand-crafted segmentation with learned discovery for COCO transferDetCon_B · COCO Mask-RCNN 1xAP bbox: 42.7 -> 42.9Improved
replace hand-crafted segmentation with learned discoveryDetCon_B · COCO FCOS-style detector 30 ep (ResNet-50)AP bbox: 45.4 -> 45.6No clear effect
pretrain with Odin instead of DINODINO · COCO FCOS-style detector 30 ep (ResNet-50)AP bbox: 44.3 -> 45.6Improved
pretrain Swin-S with Odin instead of supervisedSupervised · COCO FCOS-style detector 30 ep (Swin-S)AP bbox: 48.3 -> 50.4Improved
keep FPN when evaluating object discoveryOdin · COCO object discovery (ResNet-50)ABO instance: 41.5 -> 43.0Improved
train a ViT-B backbone with Odin instead of DINODINO · COCO object discovery (ViT-B)ABO instance: 42.7 -> 45.9Improved
fine-tune the target network instead of the online network (ablation)Odin · COCO Mask-RCNN 1xAP bbox: 42.9 -> 42.7No clear effect

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