# Structure over Depth: A Single-Block Spatio-Temporal Transformer for Multi-Entity Reasoning

arXiv: https://arxiv.org/abs/2607.23077 · HTML: https://mlsanity.com/paper/2607.23077/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 37 of its 52 references that are in the map.
- First posted on arXiv 25 Jul 2026.

## Builds on (most cited first, 30 of 37)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://mlsanity.com/paper/2010.11929/) (2020)
- [Quo Vadis, Action Recognition? A New Model and the Kinetics Dataset](https://mlsanity.com/paper/1705.07750/) (2017)
- [Non-local Neural Networks](https://mlsanity.com/paper/1711.07971/) (2017)
- [Is Space-Time Attention All You Need for Video Understanding?](https://mlsanity.com/paper/2102.05095/) (2021)
- [Spatial Temporal Graph Convolutional Networks for Skeleton-Based Action Recognition](https://mlsanity.com/paper/1801.07455/) (2018)
- [Two-Stream Convolutional Networks for Action Recognition in Videos](https://mlsanity.com/paper/1406.2199/) (2014)
- [CrossViT: Cross-Attention Multi-Scale Vision Transformer for Image Classification](https://mlsanity.com/paper/2103.14899/) (2021)
- [NTU RGB+D: A Large Scale Dataset for 3D Human Activity Analysis](https://mlsanity.com/paper/1604.02808/) (2016)
- [NTU RGB+D 120: A Large-Scale Benchmark for 3D Human Activity Understanding](https://mlsanity.com/paper/1905.04757/) (2019)
- [Channel-wise Topology Refinement Graph Convolution for Skeleton-Based Action Recognition](https://mlsanity.com/paper/2107.12213/) (2021)
- [Revisiting Skeleton-based Action Recognition](https://mlsanity.com/paper/2104.13586/) (2021)
- [Actional-Structural Graph Convolutional Networks for Skeleton-based Action Recognition](https://mlsanity.com/paper/1904.12659/) (2019)
- [Learning Salient Boundary Feature for Anchor-free Temporal Action Localization](https://mlsanity.com/paper/2103.13137/) (2021)
- [Semantics-Guided Neural Networks for Efficient Skeleton-Based Human Action Recognition](https://mlsanity.com/paper/1904.01189/) (2019)
- [Hierarchical Deep Temporal Models for Group Activity Recognition](https://mlsanity.com/paper/1607.02643/) (2016)
- [Actor-Transformers for Group Activity Recognition](https://mlsanity.com/paper/2003.12737/) (2020)
- [Learning Actor Relation Graphs for Group Activity Recognition](https://mlsanity.com/paper/1904.10117/) (2019)
- [GroupFormer: Group Activity Recognition with Clustered Spatial-Temporal Transformer](https://mlsanity.com/paper/2108.12630/) (2021)
- [SkateFormer: Skeletal-Temporal Transformer for Human Action Recognition](https://mlsanity.com/paper/2403.09508/) (2024)
- [Spatio-Temporal Dynamic Inference Network for Group Activity Recognition](https://mlsanity.com/paper/2108.11743/) (2021)
- [3Mformer: Multi-order Multi-mode Transformer for Skeletal Action Recognition](https://mlsanity.com/paper/2303.14474/) (2023)
- [Richly Activated Graph Convolutional Network for Robust Skeleton-based Action Recognition](https://mlsanity.com/paper/2008.03791/) (2020)
- [Convolutional Relational Machine for Group Activity Recognition](https://mlsanity.com/paper/1904.03308/) (2019)
- [IGFormer: Interaction Graph Transformer for Skeleton-based Human Interaction Recognition](https://mlsanity.com/paper/2207.12100/) (2022)
- [Interaction Relational Network for Mutual Action Recognition](https://mlsanity.com/paper/1910.04963/) (2019)
- [Human Activity Recognition from Wearable Sensor Data Using Self-Attention](https://mlsanity.com/paper/2003.09018/) (2020)
- [Dual-AI: Dual-path Actor Interaction Learning for Group Activity Recognition](https://mlsanity.com/paper/2204.02148/) (2022)
- [Interactive Spatiotemporal Token Attention Network for Skeleton-based General Interactive Action Recognition](https://mlsanity.com/paper/2307.07469/) (2023)
- [Focal and Global Spatial-Temporal Transformer for Skeleton-based Action Recognition](https://mlsanity.com/paper/2210.02693/) (2022)

## Abstract (arXiv)

Modeling multi-entity temporal data requires capturing dependencies across entities, time, and their interactions. Transformer-based approaches perform well but often rely on deep stacks of layers to learn these heterogeneous dependencies implicitly, increasing computational cost. We revisit this problem from a structural perspective and decompose multi-entity temporal dynamics into three interaction types: spatial interactions among entities, temporal interactions across time, and cross interactions coupling the two domains. We propose a structured spatio-temporal transformer block that explicitly models all three within a single stage. It uses parallel spatial and temporal self-attention, followed by bidirectional cross-attention, and combines the outputs through learnable gated fusion. By directly encoding these complementary views, the model reduces the need for deep stacking. We evaluate the approach on video-based group activity recognition, skeleton-based human interaction analysis, and wearable sensor-based activity recognition. Despite its simplicity, the single structured Transformer block matches or outperforms deeper architectures with only 1.76M parameters. The results suggest that depth in prior models partly compensates for implicit and entangled interaction modeling, whereas explicit factorization offers a more efficient and transparent alternative. More broadly, this work supports a structure-first design principle: expressive multi-entity temporal reasoning can emerge by exposing interaction structure rather than relying on depth.