Differences

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

--- products:sbc:vim3:npu:ksnn:ksnn-convert [2023/09/15 11:54]
sravan [Introduction]
+++ products:sbc:vim3:npu:ksnn:ksnn-convert [2024/05/08 02:16]
louis
@@ Line 3: / Line 3: @@
 Introduction and guide on how to use the KSNN conversion tool.
-===== Get convert tool =====
+===== Get the conversion tool =====
-The tool is integrated in the NPU conversion tool warehouse of VIM3.
+The conversion tool is integrated in the NPU SDK.
 ```shell
@@ Line 17: / Line 17: @@
 ```
-===== Example for Convert =====
+===== Conversion example =====
-Choose tensorflow Inception V3 as a exmple.
+Choose tensorflow MobileNet SSD as an example.
-. Get forzen model,
+. Get the frozen model,
 ```shell
 $ cd aml_npu_sdk/acuity-toolkit/python
-$ wget https://github.com/yan-wyb/models-zoo/raw/master/tensonrflow/inception/inception_v3_2016_08_28_frozen.pb
+$ wget https://github.com/yan-wyb/models-zoo/raw/master/tensorflow/mobilenet_ssd/ssd_mobilenet_v1_coco_2017_11_17.pb
 ```
@@ Line 31: / Line 31: @@
 ```shell
-$ ./convert --model-name inception \
+# uint8
+$ ./convert --model-name mobilenet_ssd \
 >           --platform tensorflow \
->           --model /home/yan/yan/Yan/models-zoo/tensorflow/inception/inception_v3_2016_08_28_frozen.pb \
+>           --model /home/yan/yan/Yan/models-zoo/tensorflow/mobilenet_ssd/ssd_mobilenet_v1_coco_2017_11_17.pb \
->           --input-size-list '299,299,3' \
+>           --input-size-list '300,300,3' \
->           --inputs input \
+>           --inputs FeatureExtractor/MobilenetV1/MobilenetV1/Conv2d_0/BatchNorm/batchnorm/mul_1 \
->           --outputs InceptionV3/Predictions/Reshape_1 \
+>           --outputs "'concat concat_1'" \
->           --mean-values '128 128 128 0.0078125' \
+>           --mean-values '127.5 127.5 127.5 0.007843137' \
 >           --quantized-dtype asymmetric_affine \
+>           --source-files ./data/dataset/dataset0.txt \
+>           --kboard VIM3 --print-level 0
+# int8
+$ ./convert --model-name mobilenet_ssd \
+>           --platform tensorflow \
+>           --model /home/yan/yan/Yan/models-zoo/tensorflow/mobilenet_ssd/ssd_mobilenet_v1_coco_2017_11_17.pb \
+>           --input-size-list '300,300,3' \
+>           --inputs FeatureExtractor/MobilenetV1/MobilenetV1/Conv2d_0/BatchNorm/batchnorm/mul_1 \
+>           --outputs "'concat concat_1'" \
+>           --mean-values '127.5 127.5 127.5 0.007843137' \
+>           --quantized-dtype dynamic_fixed_point \
+>           --qtype int8 \
+>           --source-files ./data/dataset/dataset0.txt \
+>           --kboard VIM3 --print-level 0
+# int16
+$ ./convert --model-name mobilenet_ssd \
+>           --platform tensorflow \
+>           --model /home/yan/yan/Yan/models-zoo/tensorflow/mobilenet_ssd/ssd_mobilenet_v1_coco_2017_11_17.pb \
+>           --input-size-list '300,300,3' \
+>           --inputs FeatureExtractor/MobilenetV1/MobilenetV1/Conv2d_0/BatchNorm/batchnorm/mul_1 \
+>           --outputs "'concat concat_1'" \
+>           --mean-values '127.5 127.5 127.5 0.007843137' \
+>           --quantized-dtype dynamic_fixed_point \
+>           --qtype int16 \
 >           --source-files ./data/dataset/dataset0.txt \
 >           --kboard VIM3 --print-level 0
@@ Line 45: / Line 72: @@
 During conversion, if you need to view detailed information, you can modify ''--print-level 0'' to ''--print-level 1''.
-. File generated after conversion.
+. Model file generated after conversion.
 ```shell
 $ cd aml_npu_sdk/acuity-toolkit/python
-$ ls outputs/inception/
+$ ls outputs/mobilenet_ssd/
-inception.nb  libnn_inception.so
+mobilenet_ssd.nb  libnn_mobilenet_ssd.so
 ```

Khadas Docs

User Tools

Site Tools

Differences

Page Tools