blob: df93477373ff89bf118385aae7c8beb509dadce6 (
plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
|
# candle-resnet
A candle implementation of inference using a pre-trained [ResNet](https://arxiv.org/abs/1512.03385).
This uses a classification head trained on the ImageNet dataset and returns the
probabilities for the top-5 classes.
## Running an example
```
$ cargo run --example resnet --release -- --image tiger.jpg
loaded image Tensor[dims 3, 224, 224; f32]
model built
tiger, Panthera tigris : 90.21%
tiger cat : 8.93%
lion, king of beasts, Panthera leo: 0.35%
leopard, Panthera pardus: 0.16%
jaguar, panther, Panthera onca, Felis onca: 0.09%
```
|
