วีดีโอ

You are the best for beginner like me so I encourage you to do more of this beginner friendly and playlist and example tutorial. Very much thank you, brother.

This is not v helpful. I feel like the content of the video is too general and doesn't really give insight into the choice of optimizers - maybe things like personal projects and what were the first choice (rather than just hyperparameter search with optimizers - which is fairly obvious) for different tasks would be more useful. Other than that, just trying out different things is a hard ask for large datasets and long training times (models that take a while to converge) so a better researched video (into what practitioners do) would be more helpful for a beginner to understand how people quickly select the right optimizer for the task. At the v least industry standards like super convergence with AdamW could be brought up to give people an easy choice for the start of the project ( as in this generally works). And when they ask about SGD, you could have talked about where it's generally useful like -it can converge faster if you select the right parameters, but is v prone to bad initialization unlike Adam which is why people generally start with Adam optimizers. TLDR, the video just says read documentation and try a hyperparameter search with different optimizers and parameter values and hope that it works for beginners. And that's a sentiment that's repeated through many blogs and videos and doesn't help beginners get a clue

Very nice explaination

Wilson Michael Johnson Karen Miller Amy

very descriptive. thank you. I also have one question: what if padding=1 and output_padding=1. Will the last column and last row be left intact or will it be replaced by all-0 column and row?

it was amazing

What is IOU?

nice, thank you

why not used in previous video with StandardScalar, thany u

Thanks for the video. In 3:37 minutes, you referred a video about dilation. But as this is the first video in the playlist, I am getting confused. Could you share the video link?

Thanks a Lot. Please start Transformers for NLP and Vision if possible.

NeutrAl comment

Negative comment

Positive comment

I disagree with the ease of implementation, it all depends on your own transfer learning or inference file. There are heaps of examples where you just parse items and use single CLI for both SSD and Faster-RCNN. Ultimate winner should be SSD due to it’s inference and latency plus the ability to be used commercially. Most of yolo algorithms are either AGPL or GPL licensed, which requires heaps of money to be used commercially!

Sir i want to use RCNN for number plate detection, how to do this ? Please reply

Great video. Thank you for your contributions to knowledge.

Bhaiya, i request to you please do code also so that we can learn and know more about mediapipe

This is one of the best explanation ever, thanks a lot

how to import a torch library

Bhaiya, are open cv and mediapipe relevant to learn? Is there any future of it ?? Plzz reply because I got stuck in my life. Plz show me direction 😢

The problem is that there is no code in this. I have watched a lot of TH-cam tutorials and I know that you are very good at explaining your coding which most TH-camrs take for granted. That is why I think making code focused tutorials would be better and along with you can supplement some theory

Please provide roadmap for computer vision, there's not particular path for it, like cnn,yolo, diffusion models,etc, etc

Ur linkedin link?

after rerun its getting 0 with same code

why y1 < y3 and y2 < y4?

For freshers(internship/job) should we use two column resume or one column

For freshers(internship/job) should we use two column resume or one column

very well explained

I learnt Deep Learning and Machine Learning to apply to Haeltcare Problems. Please do feature Healthcare Related Projects specifically Vision Related Tasks. If possible try to feature some less common projects which require more in depth knowledge of DL and Vision

Thanks a LOT Amazing, this is something that is overlooked EVERYWHERE. Just one request that apart from the NLP and Vision Projects please do make projects which use Anutoencoders coz thay are being overlooked. They are right now only used in Segmentation but they also can do Anomaly Detection so please do explore that topic as it is very useful and strangely extremely ignored. One more request would be to please cover Autoencoders a little bit thoretically and write it from scratch .