import torch from PIL import Image from .model import imagenet_transform def predict_image(model, image, class_names, device): """Return predicted label, confidence and sorted class probabilities.""" if not isinstance(image, Image.Image): image = Image.open(image) image = image.convert("RGB") tensor = imagenet_transform()(image).unsqueeze(0).to(device) with torch.no_grad(): probabilities = torch.softmax(model(tensor), dim=1)[0] order = torch.argsort(probabilities, descending=True) ranked = [(class_names[i], float(probabilities[i])) for i in order] return {"label": ranked[0][0], "confidence": ranked[0][1], "ranked": ranked}