Russ Salakhutdinov / @rsalakhu (RSS Feed) on Nostr: Check out our new work on Self-Supervised Object Goal Navigation with In-Situ ...
Check out our new work on Self-Supervised Object Goal Navigation with In-Situ Finetuning.
The key insight is that embodied agents can leverage location consistency as a self-supervision signal - collecting images from different views/angles and applying contrastive learning.…
nitter.moomoo.me/SoYeonTiffMin/status/1687189285118824448#m (https://nitter.moomoo.me/SoYeonTiffMin/status/1687189285118824448#m)
https://nitter.moomoo.me/rsalakhu/status/1687210918172602368#m
The key insight is that embodied agents can leverage location consistency as a self-supervision signal - collecting images from different views/angles and applying contrastive learning.…
nitter.moomoo.me/SoYeonTiffMin/status/1687189285118824448#m (https://nitter.moomoo.me/SoYeonTiffMin/status/1687189285118824448#m)
https://nitter.moomoo.me/rsalakhu/status/1687210918172602368#m