Views
No views yet
| Metric | Train | Val | Test |
|---|---|---|---|
| Top-1 Accuracy | 74.20% | 75.56% | 75.45% ⭐ |
| Top-5 Accuracy | 93.17% | — | 93.81% |
| Loss | 2.0298 | — | 1.9623 |
| Metric | Test |
|---|---|
| Top-1 Accuracy | 73.89% |
| Top-5 Accuracy | 91.86% |
1from PIL import Image
2import torch
3import torchvision.models as models
4import torchvision.transforms as transforms
5
6# Load model
7model = models.mobilenet_v3_small(weights=None, num_classes=1081)
8model.load_state_dict(torch.hub.load_state_dict_from_url(
9 'https://huggingface.co/cpoisson/plantnet300k-mobilenetv3-small/resolve/main/mobilenetv3_small_v2.pth'
10))
11model.eval()
12
13# Prepare image
14transform = transforms.Compose([
15 transforms.Resize(256),
16 transforms.CenterCrop(224),
17 transforms.ToTensor(),
18 transforms.Normalize(
19 mean=[0.485, 0.456, 0.406],
20 std=[0.229, 0.224, 0.225]
21 )
22])
23
24image = transform(Image.open('plant.jpg')).unsqueeze(0)
25
26# Predict
27with torch.no_grad():
28 logits = model(image)
29 top_k = torch.topk(logits, 5)
30 probs = torch.softmax(logits, dim=1)
31
32print(f"Top-1: {probs.max().item():.2%}")1@article{plantnet2017,
2 title={PlantNet: A Large-Scale Continuous Ecosystem for Plant Image Classification},
3 author={Cole et al.},
4 year={2017}
5}