Khadas Docs

Amazing Khadas, always amazes you!

User Tools

Site Tools

Home / products / sbc / vim4 / npu / demos / facenet
products:sbc:vim4:npu:demos:facenet

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revision Previous revision
Next revision 		Previous revision
products:sbc:vim4:npu:demos:facenet [2023/09/15 03:01]
sravan [Facenet Pytorch VIM4 Demo - 6] 		products:sbc:vim4:npu:demos:facenet [2026/04/02 02:47] (current)
nick
Line 1: Line 1:
-~~tag> NPU Facenet VIM4 Pytorch~~ +~~tag> NPU FaceNet VIM4 PyTorch~~ 
-====== Facenet Pytorch VIM4 Demo - 6 ======+ 
 +**Doc for version ddk-3.4.7.7** 
 + 
 +====== FaceNet PyTorch VIM4 Demo - 6 ======
  
 {{indexmenu_n>6}} {{indexmenu_n>6}}
 +
 +===== Introduction =====
 +
 +FaceNet is a face recognition model. It will convert a face image into a feature map. Compare the feature map between image and face database. Here are two judgment indicators, cosine similarity and Euclidean distance. The closer the cosine similarity is to 1 and the closer the Euclidean distance is to 0, the more similar is between two faces.
 +
 +Here takes **lin_1.jpg** as example. Inference results on VIM4.
 +
 +{{:products:sbc:vim4:npu:demos:facenet-demo-output.webp?400|}}
 +
 ===== Get Source Code ===== ===== Get Source Code =====
 +
 +[[gh>bubbliiiing/facenet-pytorch]]
  
 ```shell ```shell
-git clone https://github.com/bubbliiiing/facenet-pytorch.git+$ git clone https://github.com/bubbliiiing/facenet-pytorch
 ``` ```
  
Line 13: Line 27:
 ==== Build virtual environment ==== ==== Build virtual environment ====
  
-Follow Docker official docs to install Docker: [[https://docs.docker.com/engine/install/ubuntu/|Install Docker Engine on Ubuntu]].+Follow Docker official documentation to install Docker: [[https://docs.docker.com/engine/install/ubuntu/|Install Docker Engine on Ubuntu]].
  
-Get Docker.+Follow the script below to get Docker image:
  
 ```shell ```shell
-$ docker pull yanwyb/npu:v1 +docker pull numbqq/npu-vim4
-$ docker run -it --name vim4-npu1 -v $(pwd):/home/khadas/npu \ +
- -v /etc/localtime:/etc/localtime:ro \ +
- -v /etc/timezone:/etc/timezone:ro \ +
- yanwyb/npu:v1+
 ``` ```
  
-==== Get convert tool ====+==== Get Convert Tool ====
  
-Download Tool from [[gl>khadas/vim4_npu_sdk.git|Rockchip Github]].+You can find the SDK here: [[dl>products/vim4/tools/npu-sdk/]]
  
 ```shell ```shell
-$ git clone https://gitlab.com/khadas/vim4_npu_sdk.git+$ wget https://dl.khadas.com/products/vim4/tools/npu-sdk/vim4_npu_sdk-ddk-3.4.7.7-250508.tgz 
 +$ tar xvzf vim4_npu_sdk-ddk-3.4.7.7-250508.tgz 
 +$ cd vim4_npu_sdk-ddk-3.4.7.7-250508 
 +$ ls 
 +adla-toolkit-binary  adla-toolkit-binary-3.1.7.4  convert-in-docker.sh  Dockerfile  docs  README.md
 ``` ```
 +
 +  * ''docs'' - SDK documentations
 +  * ''adla-toolkit-binary/bin'' - SDK tools required for model conversion
 +  * ''adla-toolkit-binary/demo'' - Conversion examples
 +  * ''adla-toolkit-binary/python'' - KSNN conversion examples
 +
 +<WRAP important>
 +If your kernel is older than 241129, please use branch npu-ddk-1.7.5.5.
 +</WRAP>
 +
 +==== Convert ====
  
 After training model, modify ''facenet-pytorch/nets/facenet.py'' as follows. After training model, modify ''facenet-pytorch/nets/facenet.py'' as follows.
Line 50: Line 75:
 ``` ```
  
-Create a python file written as follows and run to convert model to onnx.+Create a Python file written as follows and run to convert the model to ONNX.
  
 ```python export.py ```python export.py
Line 68: Line 93:
 Enter ''vim4_npu_sdk/demo'' and modify ''convert_adla.sh'' as follows. Enter ''vim4_npu_sdk/demo'' and modify ''convert_adla.sh'' as follows.
  
-```shell convert_adla.sh+```bash convert_adla.sh
 #!/bin/bash #!/bin/bash
      
Line 86: Line 111:
         --dtypes "float32" \         --dtypes "float32" \
         --inference-input-type float32 \         --inference-input-type float32 \
- --inference-output-type float32 \+ --inference-output-type float32 \
         --quantize-dtype int8 --outdir onnx_output  \         --quantize-dtype int8 --outdir onnx_output  \
         --channel-mean-value "0,0,0,255"  \         --channel-mean-value "0,0,0,255"  \
Line 95: Line 120:
 ``` ```
  
-Run ''convert_adla.sh'' to generate VIM4 model. The converted model is ''xxx.adla'' in ''onnx_output''.+Run ''convert_adla.sh'' to generate the VIM4 model. The converted model is ''xxx.adla'' in ''onnx_output''.
  
 ```shell ```shell
Line 101: Line 126:
 ``` ```
  
-===== Run NPU =====+===== Run inference on the NPU =====
  
 ==== Get source code ==== ==== Get source code ====
Line 108: Line 133:
  
 ```shell ```shell
-$ git clone https://github.com/khadas/vim4_npu_applications.git+$ git clone https://github.com/khadas/vim4_npu_applications
 ``` ```
 +
 +<WRAP important>
 +If your kernel is older than 241129, please use version before tag ddk-3.4.7.7.
 +</WRAP>
  
 ==== Install dependencies ==== ==== Install dependencies ====
Line 122: Line 151:
 === Picture input demo === === Picture input demo ===
  
-There are two modes of this demo. One is converting face images into feature vectors and saving vectors in face library. Another is comparing input face image with faces in library and outputting Euclidean distance and cosine similarity.+There are two modes of this demo. One is converting face images into feature vectors and saving vectors in the face library. Another is comparing input face image with faces in the library and outputting Euclidean distance and cosine similarity.
  
 Put ''facenet_int8.adla'' in ''vim4_npu_applications/facenet/data/''. Put ''facenet_int8.adla'' in ''vim4_npu_applications/facenet/data/''.
Line 128: Line 157:
 ```shell ```shell
 # Compile # Compile
-$ cd vim4_npu_applications/retinaface+$ cd vim4_npu_applications/facenet
 $ mkdir build $ mkdir build
 $ cd build $ cd build
Line 135: Line 164:
  
 # Run mode 1 # Run mode 1
-$ sudo ./facenet -m ../data/facenet_int8.adla -p 1+$ ./facenet -m ../data/facenet_int8.adla -p 1
 ``` ```
  
-After running mode 1, a file named ''face_feature_lib'' will generate in ''vim4_npu_applications/facenet''. Had this file, you can run mode 2.+After running mode 1, a file named ''face_feature_lib'' will generate in ''vim4_npu_applications/facenet''. With this file generated, you can run mode 2.
  
 ```shell ```shell
 # Run mode 2 # Run mode 2
-$ sudo ./facenet -m ../data/facenet_int8.adla -p ../data/img/lin_2.jpg+$ ./facenet -m ../data/model/facenet_int8.adla -p ../data/img/lin_2.jpg
 ``` ```
  
 +Here are two comparison methods, **Euclidean distance** and **cosine similarity**.
 +
 +**Euclidean distance** is smaller, more similar between two faces.
 +
 +**Cosine similarity** is closer to 1, more similar between two faces.
Last modified: 2023/09/15 03:01 by sravan

Page Tools

Khadas Site

Khadas Community

Khadas GitHub

Khadas Downloads