Trap Attention

Monocular Depth Estimation with Manual Traps — Implementation

(Clean, centered layout for clarity and style.)

---

📑 Table of Contents

1. Paper & Poster
2. Supplemental Materials
3. Code & Checkpoint
4. Environment
5. Usage
6. Citation & If This Helps, Please Cite
7. Acknowledgement
8. Other / TODO

---

📰 Paper & Poster

Paper:
Trap Attention: Monocular Depth Estimation With Manual Traps — CVPR 2023 :contentReference[oaicite:0]{index=0}

Poster & Supplement & Related Materials:

• Poster / Virtual Session: available via CVPR 2023 Poster Page :contentReference[oaicite:1]{index=1}
• Supplemental PDF (network details / extra results): accessible here – Supplemental File :contentReference[oaicite:2]{index=2}

(If you want, you can download the poster and place under ./docs/ or ./posters/ and insert display.)

---

📦 Code & Checkpoint

• Implementation repository: ICSResearch/TrapAttention on GitHub :contentReference[oaicite:3]{index=3}
• Pretrained / checkpoint models: (as you provided earlier) Google Drive link.

Google Drive: https://drive.google.com/drive/folders/1kIXg9UP0cVWUq_7Pq20JT9_RyR-PjvkS?usp=sharing

🧰 Environment

Python 3.8• PyTorch 1.7.1 (or greater, as long as compatible) (You may install other dependencies per requirements.txt in the repo.)

▶️ Usage / Quick Start

Clone the repo, download the checkpoint, and run as follows (for example):

git clone https://github.com/ICSResearch/TrapAttention.git
cd TrapAttention

install dependencies

pip install -r requirements.txt

example usage

your_run_script.py --config configs/your_config.yaml  # adjust as needed

📝 Tip

Make sure CUDA / GPU memory is sufficient if you run high‑res inputs or large batch size.

📚 Citation – If This Code Helps, Please CiteIf you use this code (or parts of it) in your work, please cite:bibtex

@InProceedings{Ning_2023_CVPR,
  author    = {Chao Ning and Hongping Gan},
  title     = {Trap Attention: Monocular Depth Estimation With Manual Traps},
  booktitle = {Proceedings of the IEEE/CVF International Conference on Computer Vision and Pattern Recognition (CVPR)},
  year      = {2023},
  pages     = {5033–5043}
}

🙏 Acknowledgement

Thanks to the following outstanding works / libraries / communities:• Transformers / Vision‑Transformer backbones used in encoder• The community maintaining open‑source depth‑estimation toolboxes• All contributors and testers who reported bugs or improvements

⚠️ Other

/ TODO• (Optional) Add inference examples & sample outputs in /examples/• (Optional) Add visualization of depth maps (RGB → depth) in README / docs• (Optional) Add evaluation scripts and result tables