# (7/22)Computer Vision Recent Paper:ArcNet
###### tags: `paper`
[toc]

---
## Before Meeting
:::success
### Author
- Jiankang Deng
- 
- [refer](https://scholar.google.com/citations?user=Z_UoQFsAAAAJ)
- Stefanos Zafeiriou
- 
- [refer](https://scholar.google.com/citations?user=QKOH5iYAAAAJ&hl=en)
- [refer]()
- [refer](https://blog.csdn.net/u014380165/article/details/80645489)
- [refer]()

:::

[refer](https://arxiv.org/pdf/1801.07698.pdf)
[refer]()
[refer]()

---

## Recent Paper

---

### ArcFace: Additive Angular Margin Loss for Deep Face Recognition

:::success
#### Abstracion
- Deep Convolutional Neural Networks (DCNNs)
- Centre loss penalises the distance between the deep features and their corresponding class centres in the Euclidean space to achieve intra-class compactness.
- Additive Angular Margin Loss (ArcFace) to obtain highly discriminative features for face recognition.
- We show that ArcFace consistently outperforms the state-of-the-art and can be easily implemented with negligible computational overhead.
:::
:::info
#### Detail
- Introduction
- 

- There are two main lines of research to train DCNNs for face recognition.
- train a multi-class classifier which can separate different identities in the training set, such by using a softmax classifier
- learn directly an embedding, such as the triplet loss
- Additive Angular Margin Loss (ArcFace) to further improve the discriminative power of the face recognition model and to stabilise the training process
- 

- Proposed Approach
- ArcFace
- Comparison with SphereFace and CosFace
- Comparison with Other Losses
- Experiments
- Implementation Details
- Ablation Study on Losses
- Evaluation Results

:::
:::warning
#### Conclusion
- 

- we proposed an Additive Angular Margin Loss function, which can effectively enhance the discriminative power of feature embeddings learned via DCNNs for face recognition
:::
[refer]()

---
:::success
#### Abstracion
:::
:::info
#### Detail

:::
:::warning
#### Conclusion

:::
[refer]()

---