Model | Download | Download (with sample test data) | ONNX version | Opset version | Top-1 accuracy (%) | Top-5 accuracy (%) |
---|---|---|---|---|---|---|
CaffeNet | 238 MB | 244 MB | 1.1 | 3 | ||
CaffeNet | 238 MB | 244 MB | 1.1.2 | 6 | ||
CaffeNet | 238 MB | 244 MB | 1.2 | 7 | ||
CaffeNet | 238 MB | 244 MB | 1.3 | 8 | ||
CaffeNet | 238 MB | 244 MB | 1.4 | 9 | ||
CaffeNet | 233 MB | 216 MB | 1.9 | 12 | 56.27 | 79.52 |
CaffeNet-int8 | 58 MB | 39 MB | 1.9 | 12 | 56.22 | 79.52 |
CaffeNet-qdq | 59 MB | 44 MB | 1.9 | 12 | 56.25 | 79.45 |
Compared with the fp32 CaffeNet, int8 CaffeNet's Top-1 accuracy drop ratio is 0.09%, Top-5 accuracy drop ratio is 0.13% and performance improvement is 3.08x.
Note
Different preprocess methods will lead to different accuracies, the accuracy in table depends on this specific preprocess method.
The performance depends on the test hardware. Performance data here is collected with Intel® Xeon® Platinum 8280 Processor, 1s 4c per instance, CentOS Linux 8.3, data batch size is 1.
CaffeNet a variant of AlexNet. AlexNet is the name of a convolutional neural network for classification, which competed in the ImageNet Large Scale Visual Recognition Challenge in 2012.
Differences:
- not training with the relighting data-augmentation;
- the order of pooling and normalization layers is switched (in CaffeNet, pooling is done before normalization).
Caffe BVLC CaffeNet ==> Caffe2 CaffeNet ==> ONNX CaffeNet
data_0: float[1, 3, 224, 224]
prob_1: float[1, 1000]
random generated sampe test data:
- test_data_set_0
- test_data_set_1
- test_data_set_2
- test_data_set_3
- test_data_set_4
- test_data_set_5
This model is snapshot of iteration 310,000. The best validation performance during training was iteration 313,000 with validation accuracy 57.412% and loss 1.82328. This model obtains a top-1 accuracy 57.4% and a top-5 accuracy 80.4% on the validation set, using just the center crop. (Using the average of 10 crops, (4 + 1 center) * 2 mirror, should obtain a bit higher accuracy still.)
CaffeNet-int8 and CaffeNet-qdq are obtained by quantizing fp32 CaffeNet model. We use Intel® Neural Compressor with onnxruntime backend to perform quantization. View the instructions to understand how to use Intel® Neural Compressor for quantization.
onnx: 1.9.0 onnxruntime: 1.8.0
wget https://github.com/onnx/models/raw/main/vision/classification/caffenet/model/caffenet-12.onnx
Make sure to specify the appropriate dataset path in the configuration file.
bash run_tuning.sh --input_model=path/to/model \ # model path as *.onnx
--config=caffenet.yaml \
--data_path=/path/to/imagenet \
--label_path=/path/to/imagenet/label \
--output_model=path/to/save
- mengniwang95 (Intel)
- yuwenzho (Intel)
- airMeng (Intel)
- ftian1 (Intel)
- hshen14 (Intel)