hw1.py

from fastai.vision import *
from duckduckgo_search import ddg_images
from fastcore.all import *
from fastbook import *
from fastdownload import download_url
from PIL import Image
import os

def search_images(term, max_images=60):
    print(f"Searching for '{term}'")
    return L(ddg_images(term, max_results=max_images)).itemgot('image')

def download_images(dest, urls):
    print(f"Downloading {len(urls)} images")
    for url in urls:
        if not Path(dest/url).exists():
            try:
                download_url(url, dest, show_progress=False)
            except Exception as e:
                print(f"Error while downloading photo: {e}")

def resize_image(fpath, max_size):    
    ext = Path(fpath).suffix
    new_fpath = fpath.split(".")
    if len(new_fpath) > 1:
        ext = new_fpath[-1]
        new_fpath = ".".join(new_fpath[:-1])
        new_fpath = f"{new_fpath}_{max_size}.{ext}"
    else:
        new_fpath = fpath
    print(new_fpath)
    if not Path(new_fpath).exists():
        try:
            img = Image.open(fpath)
            img = img.resize((max_size, max_size))
            img.save(f"{new_fpath}")
        except:
            pass
    os.remove(fpath)

def resize_images(path, max_size):
    print(f"resizing images")
    for fpath in os.listdir(path=path):
        fpath = f"{path}/{fpath}"
        if "_400" in fpath:
            continue
        resize_image(fpath, max_size)

searches = 'football ball', 'basketball ball', 'tennis ball'
path = Path('.')

for o in searches:
    dest = (path/o)
    if not dest.exists():
        dest.mkdir(exist_ok=True, parents=True)
        download_images(dest, urls=search_images(f'{o} photo'))
    resize_images(dest, max_size=400)


dls = DataBlock(
    blocks=(ImageBlock, CategoryBlock), 
    get_items=get_image_files, 
    splitter=RandomSplitter(valid_pct=0.2, seed=42),
    get_y=parent_label,
    item_tfms=[Resize(192, method='squish')]
).dataloaders(path, bs=32)

dls.show_batch(max_n=20)

learn = vision_learner(dls, resnet18, metrics=error_rate)
learn.fine_tune(6)
learn.export()