Python Image Processing Cookbook

How to do it...

Go through the following steps to classify an image using pretrained deep neural net models with PyTorch/torchvision:

1. Implement a classify() function that accepts an input image and a pretrained model (on ImageNet), along with the model name and the ImageNet class labels, returning the classifier corresponding to the highest-predicted probability:

def classify(img, model_index, model_name, model_pred, labels):    _, index = torch.max(model_pred, 1)    model_pred, indices = torch.sort(model_pred, dim=1,\         descending=True)    percentage = torch.nn.functional.softmax(model_pred, dim=1)\        [0] * 100    draw = ImageDraw.Draw(img)    font = ImageFont.truetype(r'arial.ttf', 50)    draw.text((5, 5+model_index*50),'{}, pred: \ {},{}%'.format(model_name, ...