VGG19: Image Classification
VGGNet is a deep convolutional neural network developed by researchers from Oxford University's Visual Geometry Group and Google DeepMind. It explores the relationship between the depth of a convolutional neural network and its performance. By repeatedly stacking 33 A small convolution kernel and a 22 maximum pooling layer have successfully constructed a 16-19 layer deep convolutional neural network. Compared with the previous state-of-the-art network structure, the error rate of VGGNet is greatly reduced. In the VGGNet paper, all the small convolution kernels of 33 and the largest pooling kernel of 22 are used to continuously deepen the network structure. Improve performance.
VGG19 contains 19 hidden layers (16 convolutional layers and 3 fully connected layers)
The model can be found here
CONTENTS
Performance
| Device | SoC | Runtime | Model | Size (pixels) | Inference Time (ms) | Precision | Compute Unit | Model Download |
|---|---|---|---|---|---|---|---|---|
| AidBox QCS6490 | QCS6490 | QNN | VGG19 | 224 | 20.3 | INT8 | NPU | model download |
| AidBox QCS6490 | QCS6490 | QNN | VGG19 | 224 | - | INT16 | NPU | model download |
| AidBox QCS6490 | QCS6490 | SNPE | VGG19 | 224 | 18.4 | INT8 | NPU | model download |
| AidBox QCS6490 | QCS6490 | SNPE | VGG19 | 224 | - | INT16 | NPU | model download |
| APLUX QCS8550 | QCS8550 | QNN | VGG19 | 224 | 6.6 | INT8 | NPU | model download |
| APLUX QCS8550 | QCS8550 | QNN | VGG19 | 224 | 11.1 | INT16 | NPU | model download |
| APLUX QCS8550 | QCS8550 | SNPE | VGG19 | 224 | 4.2 | INT8 | NPU | model download |
| APLUX QCS8550 | QCS8550 | SNPE | VGG19 | 224 | 5.6 | INT16 | NPU | model download |
| AidBox GS865 | QCS8250 | SNPE | VGG19 | 224 | - | INT8 | NPU | model download |
Model Conversion
Demo models converted from AIMO(AI Model Optimizier).
The source model YOLOv5s.onnx can be found here.
The demo model conversion step on AIMO can be found blow:
| Device | SoC | Runtime | Model | Size (pixels) | Precision | Compute Unit | AIMO Conversion Steps |
|---|---|---|---|---|---|---|---|
| AidBox QCS6490 | QCS6490 | QNN | VGG19 | 640 | INT8 | NPU | View Steps |
| AidBox QCS6490 | QCS6490 | QNN | VGG19 | 640 | INT16 | NPU | View Steps |
| AidBox QCS6490 | QCS6490 | SNPE | VGG19 | 640 | INT8 | NPU | View Steps |
| AidBox QCS6490 | QCS6490 | SNPE | VGG19 | 640 | INT16 | NPU | View Steps |
| APLUX QCS8550 | QCS8550 | QNN | VGG19 | 640 | INT8 | NPU | View Steps |
| APLUX QCS8550 | QCS8550 | QNN | VGG19 | 640 | INT16 | NPU | View Steps |
| APLUX QCS8550 | QCS8550 | SNPE | VGG19 | 640 | INT8 | NPU | View Steps |
| APLUX QCS8550 | QCS8550 | SNPE | VGG19 | 640 | INT16 | NPU | View Steps |
| AidBox GS865 | QCS8250 | SNPE | VGG19 | 640 | INT8 | NPU | View Steps |
Inference
Step1: convert model
a. Prepare source model in onnx format. The source model can be found here.
b. Login AIMO and convert source model to target format. The model conversion step can follow AIMO Conversion Step in Model Conversion Sheet.
c. After conversion task done, download target model file.
Step2: install AidLite SDK
The installation guide of AidLite SDK can be found here.