Om Prakash for person Re-Identification, CNN architectures should be used, maybe combining them with RNN or LSTM sometimes help them to work better. GAN are used for generating new objects, so i can't think of any way they can be used for identifying a person in an image.
AlignedReID can also be used for person re-identification :
Please check this ,A Practical Guide to Person Re-Identification Using AlignedReID
https://medium.com/@niruhan/a-practical-guide-to-person-re-identification-using-alignedreid-7683222da644
this clearly explains everything and also provides the code and the weights of the trained model
Liu, H., Feng, J., Qi, M., Jiang, J., & Yan, S. (2017). End-to-end comparative attention networks for person re-identification. IEEE Transactions on Image Processing, 26(7), 3492-3506.
Any good feature extractor and matcher can solve this problem. there are many approaches possible- Convolutional Neural Network (CNN), CNN+RNN, CNN + attention, Generative Adversarial Network (GAN)
Thank you Shobhar Sarkar, Abhishek Gangwar and Fasil for your recommendations.
You can also consider the Camera Style Adaptation, which reduces the risk of deep learning over fittings. A complete code is available in GITHUB at below location: https://github.com/zhunzhong07/CamStyle
The code base use the python library PyTorch.
Om Prakash for person Re-Identification, CNN architectures should be used, maybe combining them with RNN or LSTM sometimes help them to work better. GAN are used for generating new objects, so i can't think of any way they can be used for identifying a person in an image.
AlignedReID can also be used for person re-identification :
Please check this ,A Practical Guide to Person Re-Identification Using AlignedReID
https://medium.com/@niruhan/a-practical-guide-to-person-re-identification-using-alignedreid-7683222da644
this clearly explains everything and also provides the code and the weights of the trained model
- Badges
- Science topic