STMicroelectronics/resnet18_pt

ResNet

Use case : Image classification

Model description

Residual Network (ResNet) introduced skip connections that enable training of very deep networks. It revolutionized deep learning by solving the degradation problem in deep networks.

ResNet features skip connections that add input to output, enabling gradient flow, with the network learning residual functions with reference to layer inputs. Batch normalization is applied after every convolution for stable training, and a bottleneck design uses 1x1-3x3-1x1 convolution patterns for efficiency.

ResNet serves as the baseline for computer vision tasks, a transfer learning source model, and is widely used for research and benchmarking.

(source: https://arxiv.org/abs/1512.03385)

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

Network information

Network Information	Value
Framework	Torch
MParams	~3.75 M
Quantization	Int8
Provenance	https://github.com/KaimingHe/deep-residual-networks
Paper	https://arxiv.org/abs/1512.03385

Network inputs / outputs

For an image resolution of NxM and P classes

Input Shape	Description
(1, N, M, 3)	Single NxM RGB image with UINT8 values between 0 and 255

Output Shape	Description
(1, P)	Per-class confidence for P classes in FLOAT32

Recommended platforms

Platform	Supported	Recommended
STM32L0	[]	[]
STM32L4	[]	[]
STM32U5	[]	[]
STM32H7	[]	[]
STM32MP1	[]	[]
STM32MP2	[]	[]
STM32N6	[x]	[x]

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)

Model	Dataset	Format	Resolution	Series	Internal RAM (KiB)	External RAM (KiB)	Weights Flash (KiB)	STEdgeAI Core version
resnet18wd4_pt_224	Imagenet	Int8	224×224×3	STM32N6	1323	0	3843.64	3.0.0

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

Model	Dataset	Format	Resolution	Board	Execution Engine	Inference time (ms)	Inf / sec	STEdgeAI Core version
resnet18wd4_pt_224	Imagenet	Int8	224×224×3	STM32N6570-DK	NPU/MCU	13.82	72.36	3.0.0

Accuracy with Imagenet dataset

Model	Format	Resolution	Top 1 Accuracy
resnet18wd4_pt	Float	224x224x3	61.35 %
resnet18wd4_pt	Int8	224x224x3	60.54 %

Model	Format	Resolution	Top 1 Accuracy
resnet18wd4_pt	Float	224x224x3	61.35 %
resnet18wd4_pt	Int8	224x224x3	60.54 %

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: ResNet — https://github.com/KaimingHe/deep-residual-networks