PreResNet

Use case : Image classification

Model description

Pre-activation ResNet (PreResNet) is a variant of ResNet that places batch normalization and activation before convolutions. This simple change improves both training dynamics and final accuracy.

PreResNet employs a pre-activation design with BN-ReLU-Conv order instead of Conv-BN-ReLU, enabling cleaner identity mappings for improved information flow through residual connections. The improved gradient flow during training results from full pre-activation applied to both main path and shortcut connections.

The architecture is well-suited for deep learning research, transfer learning with pre-activation benefits, and applications where training dynamics matter.

(source: https://arxiv.org/abs/1603.05027)

The model is quantized to int8 using ONNX Runtime and exported for efficient deployment.

Network information

Network inputs / outputs

For an image resolution of NxM and P classes

Recommended platforms

Performances

Metrics

• Measures are done with default STEdgeAI Core configuration with enabled input / output allocated option.
• All the models are trained from scratch on Imagenet dataset

Reference NPU memory footprint on Imagenet dataset (see Accuracy for details on dataset)

Reference NPU inference time on Imagenet dataset (see Accuracy for details on dataset)

Accuracy with Imagenet dataset

Dataset details: link Number of classes: 1000. To perform the quantization, we calibrated the activations with a random subset of the training set. For the sake of simplicity, the accuracy reported here was estimated on the 50000 labelled images of the validation set.

Retraining and Integration in a simple example:

Please refer to the stm32ai-modelzoo-services GitHub here

References

[1] - Dataset: Imagenet (ILSVRC 2012) — https://www.image-net.org/

[2] - Model: PreResNet — https://github.com/KaimingHe/resnet-1k-layers