Arduino Code for test Heart Rate 7 Click

 Some notes to the Fastai course

Practical Deep Learning for Coders 2022

If you intend use Pytorch and python 3.7 for adopting fastai core, maybe you got some problems and this pages will collect the error which I have search and solve some problems

Lession 1:

Bug 1: If you use windows and it cant create a folder and download data as below:

searches = 'forest','bird'

path = Path('bird_or_not')

if not path.exists():

    path.mkdir() ### Very important to make path, will not run if not make

    for o in searches:

        dest = (path/o)

        print(o)

        dest.mkdir(exist_ok=True)

        results = search_images_ddg(f'{o} photo')

        download_images(dest, urls=results[:200])

        resize_images(dest, max_size=400, dest=dest, max_workers=1)

        ### Using max_workers=1 to solve problem of multiprocessing lack support from system

Use the command path.mkdir() will permit create folder in current workbook of windows system

Use max_workers=1 will solve the problems of BrokenProcessPool. python - How to fix BrokenProcessPool: error for concurrent.futures ProcessPoolExecutor - Stack Overflow

I guess that this arose since the pytorch not support to multiprocessing well in miniconda envinronment.

For the verify_images error in the commands:

failed = verify_images(get_image_files(path))

failed.map(Path.unlink)

we can replace with:

failed = L(fns[i] for i,o in enumerate(parallel(verify_image, get_image_files(path), n_workers=1)) if not o)

failed.map(Path.unlink)