MobileNet V4

Use case : Image classification

Model description

MobileNetV4 represents the latest evolution in the MobileNet family, introducing Universal Inverted Bottleneck (UIB) blocks that unify various efficient convolution designs. It achieves state-of-the-art accuracy-efficiency trade-offs on mobile hardware.

The architecture features a flexible UIB block design accommodating various operations, optimized through Neural Architecture Search for multiple hardware platforms. It includes Mobile MQA Attention as an efficient attention mechanism for mobile deployment, providing enhanced feature extraction with improved capacity per FLOP.

MobileNetV4 is ideal for state-of-the-art mobile vision applications requiring the latest architectural improvements, though it shows quantization sensitivity (~10% drop) that should be considered for INT8 deployment.

(source: https://arxiv.org/abs/2404.10518)

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: MobileNetV4 — https://arxiv.org/abs/2404.10518