Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2603.18795 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866918397914120192 |
|---|---|
| author | Li, Yuchen Garg, Amanmeet Chaudhuri, Shalini Zhao, Rui Kessler, Garin |
| author_facet | Li, Yuchen Garg, Amanmeet Chaudhuri, Shalini Zhao, Rui Kessler, Garin |
| contents | Large Vision Language Models (LVLMs) excel at semantic understanding but struggle with fine grained spatial grounding, as the model must implicitly infer complex geometry without ever producing a spatial interpretation. We present Perceptio, a perception enhanced LVLM with 2D and 3D spatial reasoning abilities, enabled via explicit semantic segmentation tokens and depth tokens generated directly within the autoregressive sequence. Concretely, we (i) distill a VQVAE depth codebook from a strong monocular teacher to tokenize dense depth into compact sequences, and (ii) integrate SAM2 based semantic segmentation tokens and VQ-VAE depth tokens inside the LLM so the model first emits spatial tokens and then answers. To stabilize depth token generation, we introduce novel composite depth-token objectives (marker, token, and count losses) and a soft-merging technique for differentiable reconstruction. We adopt a multi-task co-training strategy across diverse datasets, letting the model learn perception tokens to tackle multiple downstream tasks. Building on InternVL, Perceptio achieves state-of-the-art performance across benchmarks: improving referring expression segmentation by +0.8/+1.4/+1.1 cIoU on RefCOCO/+/g HardBLINK spatial understanding accuracy by 10.3%, and MMBench accuracy by 1.0%, demonstrating that explicit spatial chain-of-thought materially strengthens spatial grounding in LVLMs. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_18795 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Perceptio: Perception Enhanced Vision Language Models via Spatial Token Generation Li, Yuchen Garg, Amanmeet Chaudhuri, Shalini Zhao, Rui Kessler, Garin Computer Vision and Pattern Recognition Artificial Intelligence Large Vision Language Models (LVLMs) excel at semantic understanding but struggle with fine grained spatial grounding, as the model must implicitly infer complex geometry without ever producing a spatial interpretation. We present Perceptio, a perception enhanced LVLM with 2D and 3D spatial reasoning abilities, enabled via explicit semantic segmentation tokens and depth tokens generated directly within the autoregressive sequence. Concretely, we (i) distill a VQVAE depth codebook from a strong monocular teacher to tokenize dense depth into compact sequences, and (ii) integrate SAM2 based semantic segmentation tokens and VQ-VAE depth tokens inside the LLM so the model first emits spatial tokens and then answers. To stabilize depth token generation, we introduce novel composite depth-token objectives (marker, token, and count losses) and a soft-merging technique for differentiable reconstruction. We adopt a multi-task co-training strategy across diverse datasets, letting the model learn perception tokens to tackle multiple downstream tasks. Building on InternVL, Perceptio achieves state-of-the-art performance across benchmarks: improving referring expression segmentation by +0.8/+1.4/+1.1 cIoU on RefCOCO/+/g HardBLINK spatial understanding accuracy by 10.3%, and MMBench accuracy by 1.0%, demonstrating that explicit spatial chain-of-thought materially strengthens spatial grounding in LVLMs. |
| title | Perceptio: Perception Enhanced Vision Language Models via Spatial Token Generation |
| topic | Computer Vision and Pattern Recognition Artificial Intelligence |
| url | https://arxiv.org/abs/2603.18795 |