onnxmodelzoo/densenet-12

language: en license: apache-2.0 model_name: densenet-12.onnx tags:

• validated
• vision
• classification
• densenet-121

DenseNet-121

Compared with the DenseNet-121-12, DenseNet-121-12-int8's op-1 accuracy drop ratio is 1.25% and performance improvement is 1.18x.

Note 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.

Description

DenseNet-121 is a convolutional neural network for classification.

Paper

Densely Connected Convolutional Networks

Dataset

ILSVRC2012

Source

Caffe2 DenseNet-121 ==> ONNX DenseNet

Model input and output

Input

data_0: float[1, 3, 224, 224]

Output

fc6_1: float[1, 1000, 1, 1]

Pre-processing steps

Post-processing steps

Sample test data

random generated sampe test data:

• test_data_0.npz
• test_data_1.npz
• test_data_2.npz
• test_data_set_0
• test_data_set_1
• test_data_set_2

Results/accuracy on test set

Quantization

Mask R-CNN R-50-FPN-int8 is obtained by quantizing Mask R-CNN R-50-FPN-fp32 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.

Environment

onnx: 1.9.0 onnxruntime: 1.10.0

Prepare model

wget https://github.com/onnx/models/raw/main/vision/classification/densenet-121/model/densenet-12.onnx

Model quantize

bash run_tuning.sh --input_model=path/to/model \  # model path as *.onnx
--config=densenet.yaml \
--output_model=path/to/save

References

• Intel® Neural Compressor

Contributors

• mengniwang95 (Intel)
• airMeng (Intel)
• ftian1 (Intel)
• hshen14 (Intel)

License

MIT