Khadas Docs

Amazing Khadas, always amazes you!

User Tools

Site Tools

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

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:densenet [2024/12/03 02:10]
louis 		products:sbc:vim4:npu:demos:densenet [2025/06/11 21:49] (current)
louis
Line 6: Line 6:
  
 {{indexmenu_n>3}} {{indexmenu_n>3}}
 +
 +===== Introduction =====
 +
 +Densenet_CTC is a text recognition model. It only can recognize single line text. Therefore usually, it needs to be used in conjunction with a text detection model.
 +
 +Recognition image and inference results on VIM4.
 +
 +{{:products:sbc:vim4:npu:demos:khadasteam.webp?400|}}  {{:products:sbc:vim4:npu:demos:densenet-demo-output.webp?400|}}
 +
 ===== Get the source code ===== ===== Get the source code =====
  
Line 28: Line 37:
 ==== Get the conversion tool ==== ==== Get the conversion tool ====
  
-Download The conversion tool from [[gl>khadas/vim4_npu_sdk]].+Download The conversion tool from [[gh>khadas/vim4_npu_sdk]].
  
 ```shell ```shell
-$ git clone https://gitlab.com/khadas/vim4_npu_sdk +$ git clone https://github.com/khadas/vim4_npu_sdk
-``` +
- +
-```shell +
-$ git lfs install +
-$ git lfs clone https://gitlab.com/khadas/vim4_npu_sdk.git+
 $ cd vim4_npu_sdk $ cd vim4_npu_sdk
 +$ git lfs pull
 $ ls $ ls
 adla-toolkit-binary  adla-toolkit-binary-3.1.7.4  convert-in-docker.sh  Dockerfile  docs  README.md adla-toolkit-binary  adla-toolkit-binary-3.1.7.4  convert-in-docker.sh  Dockerfile  docs  README.md
Line 47: Line 52:
  
 <WRAP important> <WRAP important>
-If your kernel is older than 241129, please use version before tag ddk-3.4.7.7.+If your kernel is older than 241129, please use branch npu-ddk-1.7.5.5
 </WRAP> </WRAP>
  
Line 79: Line 84:
 ``` ```
  
-Enter ''vim4_npu_sdk/demo'' and modify ''convert_adla.sh'' as follows. We should quantize the model to **int16** because it is very inaccurate with **int8**.+Enter ''vim4_npu_sdk/demo'' and modify ''convert_adla.sh'' as follows.
  
 ```bash convert_adla.sh ```bash convert_adla.sh
Line 100: Line 105:
         --inference-input-type float32 \         --inference-input-type float32 \
  --inference-output-type float32 \  --inference-output-type float32 \
-        --quantize-dtype int16 --outdir onnx_output  \+        --quantize-dtype int8 --outdir onnx_output  \
         --channel-mean-value "0,0,0,255"  \         --channel-mean-value "0,0,0,255"  \
         --source-file ./densenet_ctc_dataset.txt  \         --source-file ./densenet_ctc_dataset.txt  \
Line 139: Line 144:
 === Picture input demo === === Picture input demo ===
  
-Put ''densenet_ctc_int16.adla'' in ''vim4_npu_applications/densenet_ctc/data/''.+Put ''densenet_ctc_int8.adla'' in ''vim4_npu_applications/densenet_ctc/data/''.
  
 ```shell ```shell
Line 150: Line 155:
  
 # Run # Run
-$ ./densenet_ctc -m ../data/densenet_ctc_int16.adla -p ../data/KhadasTeam.png+$ ./densenet_ctc -m ../data/densenet_ctc_int8.adla -p ../data/KhadasTeam.png
 ``` ```
  
Last modified: 2024/12/03 02:10 by louis

Page Tools

Khadas Site

Khadas Community

Khadas GitHub

Khadas Downloads