CV_CNN_DOA

KO | EN

gitlite — search

#typescript #ai-agents #deepseek-harness #dsh-plugin #open-source #ai #cli #dsh #claude-code #codex #developer-tools #react

★ 52 Open GitHub ↗

A novel direction-of-arrival (DOA) estimation method based on deep complex-valued networks with sparse prior

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# CV_CNN_DOA
A novel DOA estimation method based on deep complex-valued networks with sparse prior.
  --- 
## 🚀 Latest Update

### Power Normalization for Covariance Matrices @ 2025.03
To generalize to different hardware platforms and power levels, we strongly recommend power-normalizing the covariance matrix before model training and inference, using the following formula:

#### Mathematical Definition
Given sampling covariance matrix (SCM) $\hat{R} \in \mathbb{C}^{Batch \times M \times M}$ :  
```math
\hat{R}_{\text{normalized}} = \frac{\hat{R}}{\max\left(\text{diag}(\hat{R})\right) + \epsilon}
```
--- 
## Brief
- This is the code for IEEE ICICSP 2023 paper : "Robust DOA Estimation Using Deep ComplexValued Convolutional Networks with Sparse Prior". The link is： https://ieeexplore.ieee.org/document/10390873
- If this work is helpful to you, please star this  repositorie and cite our paper：

@inproceedings{hu2023robust, 
 title={Robust doa estimation using deep complex-valued convolutional networks with sparse prior}, 
 author={Hu, Shulin and Zeng, Cao and Liu, Minti and Tao, Haihong and Zhao, Shihua and Liu, Yu}, 
 booktitle={2023 6th International Conference on Information Communication and Signal Processing (ICICSP)}, 
 pages={234--239}, 
 year={2023}, 
 organization={IEEE} 
}

CV_CNN_DOA

Explore Similar Repositories

Was this content helpful?

Recent Feedback

Download README