AI Paper Atlas

Research map / Multi-agent systems and games / Human-AI and goal-conditioned

Embodied Visual Active Learning for Semantic Segmentation

David Nilsson, Aleksis Pirinen, Erik Gärtner, Cristian Sminchisescu

AAAI 2021 · Multi-agent systems and games › Human-AI and goal-conditioned · approach: Embodied agent learning · 27 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 a jointly learned exploration and annotation policy instead of comparable pre-specified policiesRL-agent · Matterport3D test scenes, 256 actionsmIoU: 0.385 -> 0.394Improved
Use a jointly learned exploration and annotation policy instead of comparable pre-specified policiesRL-agent · Matterport3D test scenes, 100-annotation budgetmIoU: 0.485 -> 0.507Improved
Add Collect actions and propagated annotations (ablation)RL-agent · Matterport3D validation scenesmIoU: 0.415 -> 0.427Improved
Learn annotation actions jointly with exploration instead of using a heuristic annotation strategy (ablation)RL-agent · Matterport3D validation scenesmIoU: 0.411 -> 0.427Improved
Add an explicit exploration reward (ablation)RL-agent · Matterport3D validation scenesmIoU: 0.401 -> 0.427Improved
Include visual features φ_img in the policy (ablation)RL-agent · Matterport3D validation scenesmIoU: 0.378 -> 0.427Improved
Include ResNet features φ_res in the policy (ablation)RL-agent · Matterport3D validation scenesmIoU: 0.375 -> 0.427Improved
Use learnt perception actions rather than threshold-triggered actions (ablation)Space filler · Matterport3D validation scenes, space filler pathmIoU: 0.472 -> 0.454Hurt
Use randomized perception actions rather than threshold-triggered actions (ablation)Space filler · Matterport3D validation scenes, space filler pathmIoU: 0.472 -> 0.446Hurt
Use learnt perception actions rather than threshold-triggered actions (ablation)Bounce · Matterport3D validation scenes, bounce pathmIoU: 0.388 -> 0.375Hurt
Use randomized perception actions rather than threshold-triggered actions (ablation)Bounce · Matterport3D validation scenes, bounce pathmIoU: 0.388 -> 0.379Hurt
Pre-train the segmentation network on training-house views before embodied RL (ablation)RL-agent · Matterport3D validation scenesmIoU: 0.427 -> 0.461Improved

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