Book Review - Deep Learning with fastai Cookbook

[jacobusstrydom]

Great video. I love fast ai and use it for most of my image classification problems. Can't wait for the next video.

[ahmedtouzani]

Thanks sir for your amazing content we really appreciate your work

[RS-vuum]

Sreeni,
Good review. Please do more such reviews.
Thank you

[minma]

The wildcard pattern seems to a pattern encouraged by Jeremy himself. I think they do some kind of pattern matching transparently do fetch the "correct" functions.

[drforest]

Totally agree about * imports. They are a bad coding paradigm in general.

[Mai-hehv]

What are your thoughts on nnu-net. is it of any use when dealing with 256x256 patches

[blank]

Prof, I need some information on parameters of Conv2D : • kernel • kernel_size • Stride • average pooling • max pooling and others

[aldrichlat]

You should definitely do these types of videos more! Maybe you can do Goodfellow (2016) next?

It'd be nice to hear your thoughts on that "legendary" book!

[AmitChaudhary-qxmc]

thank you sir please make more lessons on fastai.

[Ajaysharma-yvzp]

Really useful review, i must try

Prof. i found one fact regarding segmentation masks, Initially our segmentation mask has unique values [0, 255] but as I resize and save masks and images in folders using cv2.write after combining all masks of microscopy cell segmentation dataset. Reading masks again changes unique values in between [0, ..., 255].

Similar I observed with other datasets, how to handle this one.

[louisraes]

Dear Sreeni, what an excellent channel you have. I have three suggestions for videos: (1) going through fastai on an own dataset (not scraped from somewhere but just, you have the images on your disk, now what?) (2) handling different types of images e.g. you have grayscale medical images, best way to go around, what if you have a mixture of magnifications, how to go about tiff, etc. (3) how to set up your directory, how to move images around, structure folders, get them on drive or paperspace etc. These are some of the things I have found the hardest and I am sure my solution is far from the most efficient one. In any case, this channel is a tremendous source.

[daily-technology]

Great Video Sreeni, just a quick question if I may ask - do the publishers or the Authors approach you directly for this kind of sponshorship offers / deals ?

[abhinandanchoubey]

As a beginner, should I start with Keras or go with fastai?