File size: 3,009 Bytes
477da44 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 |
# How to use hybrid-quantization function
## Model Source
The model used in this example come from:
https://github.com/tensorflow/models/blob/master/research/object_detection/g3doc/tf1_detection_zoo.md ssd_mobilenet_v2_coco
## Script Usage
*Usage:*
```
1. python step1.py
2. modify ssd_mobilenet_v2.quantization.cfg according to the prompt of step1.py
3. python step2.py
```
*Description:*
- The default target platform in script is 'rk3566', please modify the 'target_platform' parameter of 'rknn.config' according to the actual platform.
- If connecting board is required, please add the 'target' parameter in 'rknn.init_runtime'.
## Expected Results
This example will outputs the results of the accuracy analysis and save the result of object detection to the 'result.jpg', as follows:
```
layer_name simulator_error
entire single
cos euc cos euc
-----------------------------------------------------------------------------------------------------------------------------------------------------
[Input] FeatureExtractor/MobilenetV2/MobilenetV2/input:0 1.00000 | 0.0 1.00000 | 0.0
[exDataConvert] FeatureExtractor/MobilenetV2/MobilenetV2/input:0_int8 0.99996 | 2.0377 0.99996 | 2.0377
[Conv] Conv__350:0
[Clip] FeatureExtractor/MobilenetV2/Conv/Relu6:0 0.99998 | 9.5952 0.99998 | 9.5952
[Conv] FeatureExtractor/MobilenetV2/expanded_conv/depthwise/BatchNorm/batchnorm/add_1:0
[Clip] FeatureExtractor/MobilenetV2/expanded_conv/depthwise/Relu6:0 0.99951 | 65.269 0.99957 | 61.673
....
[Concat] concat:0_before_conv 0.99817 | 9.3381 1.00000 | 0.0317
[exDataConvert] concat:0_before_conv__int8 0.99812 | 9.4634 0.99994 | 1.6116
[Conv] concat:0_int8 0.99812 | 9.4634 0.99994 | 1.6115
[exDataConvert] concat:0 0.99812 | 9.4634 0.99994 | 1.6115
```
- Note: Different platforms, different versions of tools and drivers may have slightly different results. |