ResNet-50 is a convolutional neural network architecture that belongs to the ResNet.The key innovation in ResNet-50 is the introduction of residual blocks, which include shortcut connections (skip connections) to enable the flow of information directly from one layer to another. These shortcut connections help mitigate the vanishing gradient problem and facilitate the training of very deep networks.
| GPU | IXUCA SDK | Release | Branch |
|---|---|---|---|
| MR-V100 | 4.4.0 | 26.03 | release/26.03 |
| MR-V100 | 4.3.0 | 25.12 | release/25.12 |
Note: 请切换到与您的 SDK 版本对应的 Release 分支进行测试。请勿直接在 master 分支上运行测试，因为 master 分支可能包含与您的本地 SDK 版本不兼容的最新更改。
切换分支命令示例：
git checkout release/26.03
Pretrained model: https://download.pytorch.org/models/resnet50-0676ba61.pth
Dataset: https://www.image-net.org/download.php to download the validation dataset.
pip3 install -r ../../igie_common/requirements.txtpython3 ../../igie_common/export.py --model-name resnet50 --weight resnet50-0676ba61.pth --output resnet50.onnxexport DATASETS_DIR=/Path/to/imagenet_val/
export RUN_DIR=../../igie_common/# Accuracy
bash scripts/infer_resnet50_fp16_accuracy.sh
# Performance
bash scripts/infer_resnet50_fp16_performance.sh# Accuracy
bash scripts/infer_resnet50_int8_accuracy.sh
# Performance
bash scripts/infer_resnet50_int8_performance.sh| Model | BatchSize | Precision | FPS | Top-1(%) | Top-5(%) |
|---|---|---|---|---|---|
| ResNet50 | 32 | FP16 | 4417.29 | 76.11 | 92.85 |
| ResNet50 | 32 | INT8 | 8628.61 | 75.72 | 92.71 |