Fast.ai - From Model to Production
Warning: This is just my notes from the excellent fast.ai MOOC which you can find here
clean
To download images with Bing Image Search, sign up at Microsoft Azure for a free account. You will be given a key, which you can copy and enter in a cell as follows (replacing 'XXX' with your key and executing it):
search_images_ddg
search_images_bing
results = search_images_bing(key, 'grizzly bear')
ims = results.attrgot('contentUrl')
len(ims)
ims
dest = '../images/grizzly.jpg'
download_url(ims[0], dest)
im = Image.open(dest)
im.to_thumb(128,128)
bear_types = 'grizzly','black','teddy'
path = Path('bears')
if not path.exists():
path.mkdir()
for o in bear_types:
dest = (path/o)
dest.mkdir(exist_ok=True)
results = search_images_bing(key, f'{o} bear')
download_images(dest, urls=results.attrgot('contentUrl'))
fns = get_image_files(path)
fns
failed = verify_images(fns)
failed
failed.map(Path.unlink);
bears = DataBlock(
blocks=(ImageBlock, CategoryBlock),
get_items=get_image_files,
splitter=RandomSplitter(valid_pct=0.2, seed=42),
get_y=parent_label,
item_tfms=Resize(128))
doc(parent_label)
dls = bears.dataloaders(path)
dls.valid.show_batch(max_n=12, nrows=2)
bears = bears.new(item_tfms=Resize(128, ResizeMethod.Squish))
dls = bears.dataloaders(path)
dls.valid.show_batch(max_n=4, nrows=1)
bears = bears.new(item_tfms=Resize(128, ResizeMethod.Pad, pad_mode='zeros'))
dls = bears.dataloaders(path)
dls.valid.show_batch(max_n=4, nrows=1)
bears = bears.new(item_tfms=RandomResizedCrop(128, min_scale=0.3))
dls = bears.dataloaders(path)
dls.train.show_batch(max_n=4, nrows=1, unique=True)
bears = bears.new(item_tfms=Resize(128), batch_tfms=aug_transforms(mult=2))
dls = bears.dataloaders(path)
dls.train.show_batch(max_n=8, nrows=2, unique=True)
bears = bears.new(
item_tfms=RandomResizedCrop(224, min_scale=0.5),
batch_tfms=aug_transforms())
dls = bears.dataloaders(path)
learn = cnn_learner(dls, resnet18, metrics=error_rate)
learn.fine_tune(4)
interp = ClassificationInterpretation.from_learner(learn)
interp.plot_confusion_matrix()
interp.plot_top_losses(5, nrows=1)
cleaner = ImageClassifierCleaner(learn)
cleaner
learn.export()
path = Path()
path.ls(file_exts='.pkl')
learn_inf = load_learner(path/'export.pkl')
learn_inf.predict('../images/grizzly.jpg')
learn_inf.dls.vocab
btn_upload = widgets.FileUpload()
btn_upload
img = PILImage.create(btn_upload.data[-1])
out_pl = widgets.Output()
out_pl.clear_output()
with out_pl: display(img.to_thumb(128,128))
out_pl
pred,pred_idx,probs = learn_inf.predict(img)
lbl_pred = widgets.Label()
lbl_pred.value = f'Prediction: {pred}; Probability: {probs[pred_idx]:.04f}'
lbl_pred
btn_run = widgets.Button(description='Classify')
btn_run
def on_click_classify(change):
img = PILImage.create(btn_upload.data[-1])
out_pl.clear_output()
with out_pl: display(img.to_thumb(128,128))
pred,pred_idx,probs = learn_inf.predict(img)
lbl_pred.value = f'Prediction: {pred}; Probability: {probs[pred_idx]:.04f}'
btn_run.on_click(on_click_classify)
VBox([widgets.Label('Select your bear!'),
btn_upload, btn_run, out_pl, lbl_pred])
- Provide an example of where the bear classification model might work poorly in production, due to structural or style differences in the training data.
- Where do text models currently have a major deficiency?
- What are possible negative societal implications of text generation models?
- In situations where a model might make mistakes, and those mistakes could be harmful, what is a good alternative to automating a process?
- What kind of tabular data is deep learning particularly good at?
- What's a key downside of directly using a deep learning model for recommendation systems?
- What are the steps of the Drivetrain Approach?
- How do the steps of the Drivetrain Approach map to a recommendation system?
- Create an image recognition model using data you curate, and deploy it on the web.
- What is
DataLoaders? - What four things do we need to tell fastai to create
DataLoaders? - What does the
splitterparameter toDataBlockdo? - How do we ensure a random split always gives the same validation set?
- What letters are often used to signify the independent and dependent variables?
- What's the difference between the crop, pad, and squish resize approaches? When might you choose one over the others?
- What is data augmentation? Why is it needed?
- What is the difference between
item_tfmsandbatch_tfms? - What is a confusion matrix?
- What does
exportsave? - What is it called when we use a model for getting predictions, instead of training?
- What are IPython widgets?
- When might you want to use CPU for deployment? When might GPU be better?
- What are the downsides of deploying your app to a server, instead of to a client (or edge) device such as a phone or PC?
- What are three examples of problems that could occur when rolling out a bear warning system in practice?
- What is "out-of-domain data"?
- What is "domain shift"?
- What are the three steps in the deployment process?
- Consider how the Drivetrain Approach maps to a project or problem you're interested in.
- When might it be best to avoid certain types of data augmentation?
- For a project you're interested in applying deep learning to, consider the thought experiment "What would happen if it went really, really well?"
- Start a blog, and write your first blog post. For instance, write about what you think deep learning might be useful for in a domain you're interested in.