A root-relative pose representation with hierarchical joint offsets lets a single CNN predict multiple person poses in one pass.
Associative embedding: End-to-end learning for joint detection and grouping
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2019 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Single-Stage Multi-Person Pose Machines
A root-relative pose representation with hierarchical joint offsets lets a single CNN predict multiple person poses in one pass.