---
library_name: pytorch
---

![mobilenet_logo](resource/MobileNet.png)

MobileNetV2 introduces an efficient convolutional architecture based on inverted residual blocks and linear bottlenecks, enabling high accuracy at very low computational cost for mobile and embedded vision applications.

Original paper: [MobileNetV2: Inverted Residuals and Linear Bottlenecks](https://arxiv.org/abs/1801.04381)

# MobileNetV2

This model uses the MobileNetV2 architecture with width multiplier = 1.0, providing a balanced trade-off between accuracy and efficiency. It is optimized for low-latency, low-power inference and is commonly used for on-device image classification as well as a backbone for mobile-friendly detection and segmentation models.

Model Configuration:
- Reference implementation: [MobileNetV2](https://pytorch.org/vision/stable/models/generated/torchvision.models.mobilenet_v2.html)
- Original Weight: [MobileNet_V2_Weights.IMAGENET1K_V2](https://download.pytorch.org/models/mobilenet_v2-7ebf99e0.pth)
- Resolution: 3x224x224
- Support Cooper version:
    - Cooper SDK: [2.5.2]
    - Cooper Foundry: [2.2]

| Model | Device | Model Link |
| :-----: | :-----: | :-----: |
| MobileNetV2 | N1-655 | [Model_Link](https://huggingface.co/Ambarella/MobileNetV2/blob/main/n1-655_mobilenet_v2.bin) |
| MobileNetV2 | CV72 | [Model_Link](https://huggingface.co/Ambarella/MobileNetV2/blob/main/cv72_mobilenet_v2.bin) |
| MobileNetV2 | CV75 | [Model_Link](https://huggingface.co/Ambarella/MobileNetV2/blob/main/cv75_mobilenet_v2.bin) |