Scene Graph Generation Comparison Hub

3 papers - avg viability 6.0

Reference Surfaces

REACT++: Efficient Cross-Attention for Real-Time Scene Graph Generation(8.0)
REACT++ is a state-of-the-art scene graph generation model that balances speed and accuracy, enabling real-time applications for embodied agents.
SGG-R$^{\rm 3}$: From Next-Token Prediction to End-to-End Unbiased Scene Graph Generation(7.0)
SGG-R$^{\rm 3}$ is a structured reasoning framework that uses chain-of-thought guided fine-tuning and reinforcement learning to improve scene graph generation by addressing sparsity and bias in relation distributions.
DSFlash: Comprehensive Panoptic Scene Graph Generation in Realtime(3.0)
DSFlash is a low-latency model for comprehensive scene graph generation optimized for edge devices.