Skip to content

Commit acb4856

Browse files
EESN-WEESN-W
committed
feat: dragon q6a npu dev demos
1 parent 6fb3a32 commit acb4856

36 files changed

+2130
-2
lines changed

‎i18n/en/docusaurus-plugin-content-docs/current/common/ai/_convnext_base_qai-app-builder.mdx‎

Lines changed: 127 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,127 @@
1+
This document describes using the [QAI AppBuilder](../qai-appbuilder) Python API to perform inference with the [ConvNext-Base](https://aihub.qualcomm.com/models/convnext_base)
2+
target recognition model using Qualcomm® Hexagon™ Processor (NPU).
3+
4+
**Supported Devices**
5+
6+
| Device | SoC |
7+
| ---------- | ------- |
8+
| Dragon Q6A | QCS6490 |
9+
10+
## Install QAI AppBuilder
11+
12+
:::tip
13+
14+
1. Please install QAI AppBuilder according to [**QAI AppBuilder Installation Guide**](../qai-appbuilder#installation-methods).
15+
16+
2. Please configure QAIRT environment variables according to [**Configure QAIRT Environment Variables**](../qai-appbuilder#configure-qairt-environment-variables).
17+
:::
18+
19+
## Run Example
20+
21+
### Install Dependencies
22+
23+
<NewCodeBlock tip="Device" type="device">
24+
25+
```bash
26+
pip3 install requests tqdm qai-hub py3_wget opencv-python torch torchvision
27+
```
28+
29+
</NewCodeBlock>
30+
31+
### Run Script
32+
33+
- Navigate to the example directory
34+
35+
<Tabs>
36+
37+
<TabItem value="QCS6490">
38+
39+
<NewCodeBlock tip="Device" type="device">
40+
41+
```bash
42+
cd ai-engine-direct-helper/samples/linux/python
43+
```
44+
45+
</NewCodeBlock>
46+
47+
</TabItem>
48+
49+
</Tabs>
50+
51+
- Prepare input image, using the following image as input example
52+
53+
{" "}
54+
55+
<div style={{ textAlign: "center" }}>
56+
<img src="/en/img/dragon/q6a/test_image.webp" style={{ width: "65%" }} />
57+
input image
58+
</div>
59+
60+
- Execute inference
61+
62+
<NewCodeBlock tip="Device" type="device">
63+
64+
```bash
65+
python3 convnext_base/convnext_base.py
66+
```
67+
68+
</NewCodeBlock>
69+
70+
```bash
71+
$ python3 convnext_base/convnext_base.py
72+
Current file directory: /mnt/ssd/qualcomm/701/zzf_fork/ai-engine-direct-helper/samples/linux/python/convnext_base
73+
0.0ms [WARNING] <W> Initializing HtpProvider
74+
75+
/prj/qct/webtech_scratch20/mlg_user_admin/qaisw_source_repo/rel/qairt-2.37.1/point_release/SNPE_SRC/avante-tools/prebuilt/dsp/hexagon-sdk-5.4.0/ipc/fastrpc/rpcmem/src/rpcmem_android.c:38:dummy call to rpcmem_init, rpcmem APIs will be used from libxdsprpc
76+
0.0ms [WARNING] <W> This META does not have Alloc2 Support
77+
78+
0.0ms [WARNING] <W> This META does not have Alloc2 Support
79+
80+
0.0ms [WARNING] <W> This META does not have Alloc2 Support
81+
82+
0.0ms [WARNING] <W> This META does not have Alloc2 Support
83+
84+
193.9ms [WARNING] Time: Read model file to memory. 71.69
85+
86+
0.0ms [WARNING] <W> This META does not have Alloc2 Support
87+
88+
0.0ms [WARNING] <W> This META does not have Alloc2 Support
89+
90+
0.0ms [WARNING] <W> This META does not have Alloc2 Support
91+
92+
0.0ms [WARNING] <W> This META does not have Alloc2 Support
93+
94+
0.0ms [WARNING] <W> This META does not have Alloc2 Support
95+
96+
0.0ms [WARNING] <W> This META does not have Alloc2 Support
97+
98+
0.0ms [WARNING] <W> This META does not have Alloc2 Support
99+
100+
269.5ms [WARNING] Time: contextCreateFromBinary. 75.47
101+
102+
269.5ms [WARNING] Time: UnmapViewOfFile. 0.00
103+
104+
270.9ms [WARNING] Time: model_initialize convnext_base 270.85
105+
106+
348.3ms [WARNING] Time: model_inference convnext_base 30.17
107+
108+
Top 5 predictions for image:
109+
110+
Samoyed 0.8726790547
111+
Pomeranian 0.0192712061
112+
keeshond 0.0150661934
113+
Japanese spaniel 0.0032834315
114+
Eskimo dog 0.0020474677
115+
0.0ms [WARNING] <W> This META does not have Alloc2 Support
116+
117+
0.0ms [WARNING] <W> This META does not have Alloc2 Support
118+
119+
0.0ms [WARNING] <W> This META does not have Alloc2 Support
120+
121+
0.0ms [WARNING] <W> This META does not have Alloc2 Support
122+
123+
/prj/qct/webtech_scratch20/mlg_user_admin/qaisw_source_repo/rel/qairt-2.37.1/point_release/SNPE_SRC/avante-tools/prebuilt/dsp/hexagon-sdk-5.4.0/ipc/fastrpc/rpcmem/src/rpcmem_android.c:42:dummy call to rpcmem_deinit, rpcmem APIs will be used from libxdsprpc
124+
365.5ms [WARNING] Time: model_destroy convnext_base 14.44
125+
```
126+
127+
The printed results show that `Samoyed` has the highest confidence, which matches the input image content.

‎i18n/en/docusaurus-plugin-content-docs/current/common/ai/_convnext_tiny_qai-app-builder.mdx‎

Lines changed: 263 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,263 @@
1+
This document describes using the [QAI AppBuilder](../qai-appbuilder) Python API to perform inference with the [ConvNext-Tiny](https://aihub.qualcomm.com/models/convnext_tiny)
2+
target recognition model using Qualcomm® Hexagon™ Processor (NPU).
3+
4+
**Supported Devices**
5+
6+
| Device | SoC |
7+
| ---------- | ------- |
8+
| Dragon Q6A | QCS6490 |
9+
10+
## Install QAI AppBuilder
11+
12+
:::tip
13+
14+
1. Please install QAI AppBuilder according to [**QAI AppBuilder Installation Guide**](../qai-appbuilder).
15+
16+
2. Please configure QAIRT environment variables according to [**Configure QAIRT Environment Variables**](../qai-appbuilder).
17+
:::
18+
19+
## Run Example
20+
21+
### Install Dependencies
22+
23+
<NewCodeBlock tip="Device" type="device">
24+
25+
```bash
26+
pip3 install requests tqdm qai-hub py3_wget opencv-python torch torchvision
27+
```
28+
29+
</NewCodeBlock>
30+
31+
### Run Script
32+
33+
- Navigate to the example directory
34+
35+
<Tabs>
36+
37+
<TabItem value="QCS6490">
38+
39+
<NewCodeBlock tip="Device" type="device">
40+
41+
```bash
42+
cd ai-engine-direct-helper/samples/linux/python
43+
```
44+
45+
</NewCodeBlock>
46+
47+
</TabItem>
48+
49+
</Tabs>
50+
51+
- Prepare input image, using the following image as input example
52+
53+
{" "}
54+
55+
<div style={{ textAlign: "center" }}>
56+
<img src="/en/img/dragon/q6a/test_image.webp" style={{ width: "65%" }} />
57+
input image
58+
</div>
59+
60+
- Execute inference
61+
62+
<NewCodeBlock tip="Device" type="device">
63+
64+
```bash
65+
python3 convnext_tiny/convnext_tiny.py
66+
```
67+
68+
</NewCodeBlock>
69+
70+
```bash
71+
$ python3 convnext_tiny/convnext_tiny.py
72+
Current file directory: /mnt/ssd/qualcomm/701/zzf_fork/ai-engine-direct-helper/samples/linux/python/convnext_tiny
73+
0.0ms [WARNING] <W> Initializing HtpProvider
74+
75+
/prj/qct/webtech_scratch20/mlg_user_admin/qaisw_source_repo/rel/qairt-2.37.1/point_release/SNPE_SRC/avante-tools/prebuilt/dsp/hexagon-sdk-5.4.0/ipc/fastrpc/rpcmem/src/rpcmem_android.c:38:dummy call to rpcmem_init, rpcmem APIs will be used from libxdsprpc
76+
0.0ms [WARNING] <W> This META does not have Alloc2 Support
77+
78+
0.0ms [WARNING] <W> This META does not have Alloc2 Support
79+
80+
0.0ms [WARNING] <W> This META does not have Alloc2 Support
81+
82+
0.0ms [WARNING] <W> This META does not have Alloc2 Support
83+
84+
132.1ms [WARNING] Time: Read model file to memory. 15.88
85+
86+
0.0ms [WARNING] <W> This META does not have Alloc2 Support
87+
88+
0.0ms [WARNING] <W> This META does not have Alloc2 Support
89+
90+
0.0ms [WARNING] <W> This META does not have Alloc2 Support
91+
92+
0.0ms [WARNING] <W> This META does not have Alloc2 Support
93+
94+
0.0ms [WARNING] <W> This META does not have Alloc2 Support
95+
96+
0.0ms [WARNING] <W> This META does not have Alloc2 Support
97+
98+
0.0ms [WARNING] <W> This META does not have Alloc2 Support
99+
100+
176.1ms [WARNING] Time: contextCreateFromBinary. 44.01
101+
102+
176.2ms [WARNING] Time: UnmapViewOfFile. 0.00
103+
104+
176.6ms [WARNING] Time: model_initialize convnext_tiny 176.53
105+
106+
245.2ms [WARNING] Time: model_inference convnext_tiny 11.95
107+
108+
Top 5 predictions for image:
109+
110+
Samoyed 0.637629509
111+
Pomeranian 0.1382825226
112+
keeshond 0.0217001885
113+
Arctic fox 0.0138142202
114+
white wolf 0.0115938596
115+
0.0ms [WARNING] <W> This META does not have Alloc2 Support
116+
117+
0.0ms [WARNING] <W> This META does not have Alloc2 Support
118+
119+
0.0ms [WARNING] <W> This META does not have Alloc2 Support
120+
121+
0.0ms [WARNING] <W> This META does not have Alloc2 Support
122+
123+
/prj/qct/webtech_scratch20/mlg_user_admin/qaisw_source_repo/rel/qairt-2.37.1/point_release/SNPE_SRC/avante-tools/prebuilt/dsp/hexagon-sdk-5.4.0/ipc/fastrpc/rpcmem/src/rpcmem_android.c:42:dummy call to rpcmem_deinit, rpcmem APIs will be used from libxdsprpc
124+
263.1ms [WARNING] Time: model_destroy convnext_tiny 15.29
125+
```
126+
127+
The printed results show that `Samoyed` has the highest confidence, which matches the input image content.
128+
129+
---
130+
131+
## English Translation (For Review)
132+
133+
This document describes using the [QAI AppBuilder](../qai-appbuilder) Python API to perform inference with the [ConvNext-Tiny](https://aihub.qualcomm.com/models/convnext_tiny)
134+
target recognition model using Qualcomm® Hexagon™ Processor (NPU).
135+
136+
**Supported Devices**
137+
138+
| Device | SoC |
139+
| ---------- | ------- |
140+
| Dragon Q6A | QCS6490 |
141+
142+
## Install QAI AppBuilder
143+
144+
:::tip
145+
146+
1. Please install QAI AppBuilder according to [**QAI AppBuilder Installation Guide**](../qai-appbuilder#installation-methods).
147+
148+
2. Please configure QAIRT environment variables according to [**Configure QAIRT Environment Variables**](../qai-appbuilder#configure-qairt-environment-variables).
149+
:::
150+
151+
## Run Example
152+
153+
### Install Dependencies
154+
155+
<NewCodeBlock tip="Device" type="device">
156+
157+
```bash
158+
pip3 install requests tqdm qai-hub py3_wget opencv-python torch torchvision
159+
```
160+
161+
</NewCodeBlock>
162+
163+
### Run Script
164+
165+
- Navigate to the example directory
166+
167+
<Tabs>
168+
169+
<TabItem value="QCS6490">
170+
171+
<NewCodeBlock tip="Device" type="device">
172+
173+
```bash
174+
cd ai-engine-direct-helper/samples/linux/python
175+
```
176+
177+
</NewCodeBlock>
178+
179+
</TabItem>
180+
181+
</Tabs>
182+
183+
- Prepare input image, using the following image as input example
184+
185+
{" "}
186+
187+
<div style={{ textAlign: "center" }}>
188+
<img src="/en/img/dragon/q6a/test_image.webp" style={{ width: "65%" }} />
189+
input image
190+
</div>
191+
192+
- Execute inference
193+
194+
<NewCodeBlock tip="Device" type="device">
195+
196+
```bash
197+
python3 convnext_tiny/convnext_tiny.py
198+
```
199+
200+
</NewCodeBlock>
201+
202+
```bash
203+
$ python3 convnext_tiny/convnext_tiny.py
204+
Current file directory: /mnt/ssd/qualcomm/701/zzf_fork/ai-engine-direct-helper/samples/linux/python/convnext_tiny
205+
0.0ms [WARNING] <W> Initializing HtpProvider
206+
207+
/prj/qct/webtech_scratch20/mlg_user_admin/qaisw_source_repo/rel/qairt-2.37.1/point_release/SNPE_SRC/avante-tools/prebuilt/dsp/hexagon-sdk-5.4.0/ipc/fastrpc/rpcmem/src/rpcmem_android.c:38:dummy call to rpcmem_init, rpcmem APIs will be used from libxdsprpc
208+
0.0ms [WARNING] <W> This META does not have Alloc2 Support
209+
210+
0.0ms [WARNING] <W> This META does not have Alloc2 Support
211+
212+
0.0ms [WARNING] <W> This META does not have Alloc2 Support
213+
214+
0.0ms [WARNING] <W> This META does not have Alloc2 Support
215+
216+
132.1ms [WARNING] Time: Read model file to memory. 15.88
217+
218+
0.0ms [WARNING] <W> This META does not have Alloc2 Support
219+
220+
0.0ms [WARNING] <W> This META does not have Alloc2 Support
221+
222+
0.0ms [WARNING] <W> This META does not have Alloc2 Support
223+
224+
0.0ms [WARNING] <W> This META does not have Alloc2 Support
225+
226+
0.0ms [WARNING] <W> This META does not have Alloc2 Support
227+
228+
0.0ms [WARNING] <W> This META does not have Alloc2 Support
229+
230+
0.0ms [WARNING] <W> This META does not have Alloc2 Support
231+
232+
176.1ms [WARNING] Time: contextCreateFromBinary. 44.01
233+
234+
176.2ms [WARNING] Time: UnmapViewOfFile. 0.00
235+
236+
176.6ms [WARNING] Time: model_initialize convnext_tiny 176.53
237+
238+
245.2ms [WARNING] Time: model_inference convnext_tiny 11.95
239+
240+
Top 5 predictions for image:
241+
242+
Samoyed 0.637629509
243+
Pomeranian 0.1382825226
244+
keeshond 0.0217001885
245+
Arctic fox 0.0138142202
246+
white wolf 0.0115938596
247+
0.0ms [WARNING] <W> This META does not have Alloc2 Support
248+
249+
0.0ms [WARNING] <W> This META does not have Alloc2 Support
250+
251+
0.0ms [WARNING] <W> This META does not have Alloc2 Support
252+
253+
0.0ms [WARNING] <W> This META does not have Alloc2 Support
254+
255+
/prj/qct/webtech_scratch20/mlg_user_admin/qaisw_source_repo/rel/qairt-2.37.1/point_release/SNPE_SRC/avante-tools/prebuilt/dsp/hexagon-sdk-5.4.0/ipc/fastrpc/rpcmem/src/rpcmem_android.c:42:dummy call to rpcmem_deinit, rpcmem APIs will be used from libxdsprpc
256+
263.1ms [WARNING] Time: model_destroy convnext_tiny 15.29
257+
```
258+
259+
The printed results show that `Samoyed` has the highest confidence, which matches the input image content.
260+
261+
```
262+
263+
```

0 commit comments

Comments
 (0)