Top AI Repos — open-source AI, indexed and scored
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
The official repo for "SpatialBot: Precise Spatial Understanding with Vision Language Models.
| Date | Stars |
|---|---|
| 2026-07-31 | 349 |
| 2026-08-18 | 350 |
| 2026-08-19 | 351 |
| 2026-08-30 | 352 |
| 2026-09-02 | 353 |
| 2026-09-09 | 354 |
| 2026-09-13 | 354 |
| 2026-09-20 | 355 |
Today
+1 stars today
This week
+1 stars this week
This month
+4 stars this month
Momentum
0.0
growth rate 0.28%/day
<h1 align = "center">
SpatialBot (ICRA 2025)
</h1>
<p align="center">
<a href="https://ieeexplore.ieee.org/abstract/document/11128671">
<img alt="ICRA Paper" src="http://img.shields.io/badge/Paper-ICRA-B31B1B.svg">
</a>
<a href="https://arxiv.org/abs/2406.13642">
<img alt="arXiv Paper" src="http://img.shields.io/badge/Paper-arXiv full length-orange">
</a>
<a href="https://huggingface.co/RussRobin/SpatialBot-3B">
<img alt="Model SpatialBot-3B" src="https://img.shields.io/badge/🤗%20Model-SpatialBot--3B-yellow">
</a>
<a href="https://huggingface.co/datasets/RussRobin/SpatialQA">
<img alt="Data" src="https://img.shields.io/badge/🤗%20Data-SpatialQA-green">
</a>
<a href="https://huggingface.co/datasets/RussRobin/SpatialBench">
<img alt="Benchmark" src="https://img.shields.io/badge/🤗%20Benchmark-SpatialBench-blue">
</a>
<a href="https://huggingface.co/datasets/RussRobin/SpatialQA-E">
<img alt="Model SpatialQA-Embodiment" src="https://img.shields.io/badge/🤗%20Embodiment Dataset-SpatialQA--E-indigo">
</a>
<a href="https://mp.weixin.qq.com/s/X1iqkkEMsop9DGCY08AfCw">
<img alt="News in Chinese" src="https://img.shields.io/badge/📰%20News_in_Chinese-机器之心-purple">
</a>
</p>
[//]: # (<a href="https://huggingface.co/datasets/RussRobin/SpatialQA">)
[//]: # ( <img alt="Dataset" src="https://img.shields.io/badge/🤗%20Dataset-SpatialQA-yellow">)
[//]: # ( </a>)
This is the official repo for ICRA 2025 paper: "SpatialBot: Precise Spatial Understanding with Vision Language Models".
SJTU, Stanford, BAAI, PKU, Oxford, SEU
<!--  -->
Model: [🤗3B model in HF](https://huggingface.co/RussRobin/SpatialBot-3B) | [🤗3B ckpt in HF](https://huggingface.co/RussRobin/SpatialBot-3B-LoRA) | [🤖3B model in wisemodel](https://wisemodel.cn/models/RussellRobin/SpatialBot-3B) | [🤖3B ckpt in wisemodel](https://wisemodel.cn/models/RussellRobin/SpatialBot-3B-LoRA)
Training set SpatialQA: [🤗SpatialQA](https://huggingface.co/datasets/RussRobin/SpatialQA)
Embodiment training set SpatialQA-E: [🤗SpatialQA-E](https://huggingface.co/datasets/RussRobin/SpatialQA-E)
Benchmark: [🤗SpatialBench](https://huggingface.co/datasets/RussRobin/SpatialBench) | [🤖SpatialBench in wisemodel](https://wisemodel.cn/datasets/RussellRobin/SpatialBench/file)
Short Paper: [📒ICRA](https://ieeexplore.ieee.org/abstract/document/11128671)
Full Length Paper: [📃General VQA + Embodiment arXiv](https://arxiv.org/abs/2406.13642)
Embodiment Videos Preview: [⚙ SpatialBot in embodiment](https://drive.google.com/drive/folders/1WBt5M0h2Z8k_ohPVUVEwIcCDaLxF9yv9?usp=sharing)
## 🚀 Quickstart
1. Install dependencies first:
```
pip install torch transformers accelerate pillow numpy
```
2. Download [SpatialBot-3B](https://huggingface.co/RussRobin/SpatialBot-3B).
Users in mainland China may want to download HF model from [HF mirror site](https://hf-mirror.com/) and change ```model_name``` to local path of SpatialBot-3B folder.
3. Run the model:
```
import torch
import transformers
from transformers import AutoModelForCausalLM, AutoTokenizer
from PIL import Image
import warnings
import numpy as np
# disable some warnings
transformers.logging.set_verbosity_error()
transformers.logging.disable_progress_bar()
warnings.filterwarnings('ignore')
# set device
device = 'cuda' # or cpu
model_name = 'RussRobin/SpatialBot-3B'
offset_bos = 0
# create model
model = AutoModelForCausalLM.from_pretrained(
model_name,
torch_dtype=torch.float16, # float32 for cpu
device_map='auto',
trust_remote_code=True)
tokenizer = AutoTokenizer.from_pretrained(
model_name,
trust_remote_code=True)
# text prompt
prompt = 'What is the depth value of point <0.5,0.2>? Answer directly from depth map.'
text = f"A chat between a curious user and an artificial intelligence assistant. The assistant gives helpful, detailed, and polite answers to the useExcerpt of 14,186 characters
Read on GitHubWould you bet a product on this? Bounded 0–100 and slow moving.
matched fp:108d88d7b6fa8e43, desc:vision-language