Views
No views yet
1conda create -n lingbot-map python=3.10 -y
2conda activate lingbot-mappip install torch==2.9.1 torchvision==0.24.1 --index-url https://download.pytorch.org/whl/cu128For other CUDA versions, see PyTorch Get Started.
pip install -e .1# CUDA 12.8 + PyTorch 2.9
2pip install flashinfer-python -i https://flashinfer.ai/whl/cu128/torch2.9/For other CUDA/PyTorch combinations, see FlashInfer installation. If FlashInfer is not installed, the model falls back to SDPA (PyTorch native attention) via--use_sdpa.
pip install -e ".[vis]"| Model Name | Huggingface Repository | ModelScope Repository | Description |
|---|---|---|---|
| lingbot-map | robbyant/lingbot-map | Robbyant/lingbot-map | Base model checkpoint (4.63 GB) |
1python demo.py --model_path /path/to/checkpoint.pt \
2 --image_folder /path/to/images/1python demo.py --model_path /path/to/checkpoint.pt \
2 --video_path video.mp4 --fps 10--keyframe_interval to reduce KV cache memory by only keeping every N-th frame as a keyframe. Non-keyframe frames still produce predictions but are not stored in the cache. This is useful for long sequences
which excesses 320 frames.1python demo.py --model_path /path/to/checkpoint.pt \
2 --image_folder /path/to/images/ --keyframe_interval 61python demo.py --model_path /path/to/checkpoint.pt \
2 --video_path video.mp4 --fps 10 \
3 --mode windowed --window_size 641# Install onnxruntime (required)
2pip install onnxruntime # CPU
3# or
4pip install onnxruntime-gpu # GPU (faster for large image sets)skyseg.onnx) will be automatically downloaded from HuggingFace on first use.1python demo.py --model_path /path/to/checkpoint.pt \
2 --image_folder /path/to/images/ --mask_sky<image_folder>_sky_masks/ so subsequent runs skip regeneration.1python demo.py --model_path /path/to/checkpoint.pt \
2 --image_folder /path/to/images/ --use_sdpa1@article{chen2026geometric,
2 title={Geometric Context Transformer for Streaming 3D Reconstruction},
3 author={Chen, Lin-Zhuo and Gao, Jian and Chen, Yihang and Cheng, Ka Leong and Sun, Yipengjing and Hu, Liangxiao and Xue, Nan and Zhu, Xing and Shen, Yujun and Yao, Yao and Xu, Yinghao},
4 journal={arXiv preprint arXiv:2604.14141},
5 year={2026}
6}