nkkbr/ViCA
This is the official implementation of ViCA2 (Visuospatial Cognitive Assistant 2), a multimodal large language model designed for advanced visuospatial reasoning. The repository also provides training scripts for the original ViCA model.
This project offers tools to build advanced AI models that can understand and reason about spatial relationships in images and videos, such as object positions, distances, and event sequences. It takes visual data (images/videos) and text queries, outputting detailed spatial answers. Scientists, robotics engineers, and developers creating AI for navigation or video analysis would find this useful.
Use this if you need to train or fine-tune AI models for complex visuospatial tasks like understanding object layouts, temporal order in videos, or geometric measurements, where off-the-shelf models struggle.
Not ideal if your primary goal is general image captioning or basic visual question-answering that doesn't require fine-grained spatial reasoning or multi-step thought processes.
Stars
12
Forks
—
Language
Python
License
—
Category
Last pushed
Dec 15, 2025
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/transformers/nkkbr/ViCA"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
KimMeen/Time-LLM
[ICLR 2024] Official implementation of " 🦙 Time-LLM: Time Series Forecasting by Reprogramming...
om-ai-lab/VLM-R1
Solve Visual Understanding with Reinforced VLMs
bytedance/SALMONN
SALMONN family: A suite of advanced multi-modal LLMs
NVlabs/OmniVinci
OmniVinci is an omni-modal LLM for joint understanding of vision, audio, and language.
fixie-ai/ultravox
A fast multimodal LLM for real-time voice