Transformers
Safetensors
trace_anything
feature-extraction
trace-anything
4d
trajectory-field
video
video-to-3d
custom_code
Instructions to use bigshanedogg/Trace-Anything with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use bigshanedogg/Trace-Anything with Transformers:
# pip install -U transformers accelerate # Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("bigshanedogg/Trace-Anything", trust_remote_code=True, device_map="auto") - Notebooks
- Google Colab
- Kaggle
Download NOTICE from bigshanedogg/Trace-Anything: direct link, hf CLI and curl.
- Browser
- Download file 2.03 kB
-
https://huggingface.co/bigshanedogg/Trace-Anything/resolve/main/NOTICE
- Command line
-
hf download hf://bigshanedogg/Trace-Anything/NOTICE
-
curl -L -o NOTICE https://huggingface.co/bigshanedogg/Trace-Anything/resolve/main/NOTICE
2.03 kB
| Trace-Anything -- unofficial transformers-format package (trust_remote_code) | |
| Packaged by bigshanedogg, 2026. NOT an official release of the original authors. | |
| Components | |
| ---------- | |
| 1. Trace Anything weights (trace_anything.pt) | |
| Source: https://huggingface.co/depth-anything/trace-anything | |
| Copyright: © 2025 Bytedance Ltd. and/or its affiliates (Xinhang Liu, Yuxi Xiao, Donny Y. Chen, Jiashi Feng, Yu-Wing Tai, Chi-Keung Tang, Bingyi Kang) | |
| License: CC-BY-NC-4.0 (non-commercial) -- LICENSE, LICENSE_MODEL.txt | |
| 2. Trace Anything code (trace_anything/, vendored into ta_*.py; scripts/infer.py _load_images in the image processor, configs/eval.yaml model.net in config.json) | |
| Source: https://github.com/ByteDance-Seed/TraceAnything @ 54677b5e7bf11510c2e8c917a509988ad379f8eb | |
| Copyright: Copyright (c) 2025 Bytedance Ltd. and/or its affiliates | |
| License: Apache-2.0 -- LICENSE.code | |
| Changes made in this package | |
| ---------------------------- | |
| - ta_*.py: the official files concatenated into flat files; package-internal imports replaced by relative imports. No computation changed. | |
| - configuration_ / modeling_ / image_processing_ / processing_trace_anything.py: new transformers wrappers; the model runs TraceAnything.forward; the image processor is infer.py _load_images. | |
| - Local head left out: configs/eval.yaml sets head_args.with_local_head, but the released checkpoint holds no ds_head_local weights (infer.py's strict=False load leaves that head randomly initialised, so its *_local outputs carry no trained signal). config.json sets with_local_head false; every other output is unchanged. | |
| - model.safetensors: the trace_anything.pt state dict, converted by save_pretrained, names unchanged, every tensor checked equal (strict load). | |
| License files | |
| ------------- | |
| - LICENSE: Creative Commons Attribution-NonCommercial 4.0 International (weights) | |
| - LICENSE_MODEL.txt: the weights' license notice, copy from ByteDance-Seed/TraceAnything | |
| - LICENSE.code: Apache License 2.0, copy from ByteDance-Seed/TraceAnything (code) | |