v0.46.1
Browse filesSee https://github.com/quic/ai-hub-models/releases/v0.46.1 for changelog.
- Inception-v3_float.dlc +0 -3
- Inception-v3_float.onnx.zip +0 -3
- Inception-v3_float.tflite +0 -3
- Inception-v3_w8a8.dlc +0 -3
- Inception-v3_w8a8.onnx.zip +0 -3
- Inception-v3_w8a8.tflite +0 -3
- README.md +113 -262
- tool-versions.yaml +0 -4
Inception-v3_float.dlc
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:1684919a87b66fe0d268b9e7ca39269eceb6d9874a0386022fc1ebf745636ddd
|
| 3 |
-
size 95527100
|
|
|
|
|
|
|
|
|
|
|
|
Inception-v3_float.onnx.zip
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:9922615c90115de8c22e4fa2d3887bdfbe06c2fb4a4416bd45d031ce2dacc5a6
|
| 3 |
-
size 88766229
|
|
|
|
|
|
|
|
|
|
|
|
Inception-v3_float.tflite
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:11f8bc6ee2847bc3c1258e52cd7afd6fe65f652e9854a400a2e8ecf3ceb9ff2e
|
| 3 |
-
size 95312692
|
|
|
|
|
|
|
|
|
|
|
|
Inception-v3_w8a8.dlc
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:eb6a55bdc2cfc912a265a5234ae700fbd0a89d4e00d33b7429c885a23877b1c5
|
| 3 |
-
size 25105396
|
|
|
|
|
|
|
|
|
|
|
|
Inception-v3_w8a8.onnx.zip
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:6706289d3a8a05f86ead63590313f17c56b6b237e926663a3fd8da8db3edc56e
|
| 3 |
-
size 41143826
|
|
|
|
|
|
|
|
|
|
|
|
Inception-v3_w8a8.tflite
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:b448c50c76b811c1fa69ebb077e15edb741fd66a6836153ac5de3010959f4fde
|
| 3 |
-
size 24447824
|
|
|
|
|
|
|
|
|
|
|
|
README.md
CHANGED
|
@@ -10,280 +10,131 @@ pipeline_tag: image-classification
|
|
| 10 |
|
| 11 |
|
| 12 |
|
| 13 |
-
# Inception-v3: Optimized for
|
| 14 |
-
## Imagenet classifier and general purpose backbone
|
| 15 |
-
|
| 16 |
|
| 17 |
InceptionNetV3 is a machine learning model that can classify images from the Imagenet dataset. It can also be used as a backbone in building more complex models for specific use cases.
|
| 18 |
|
| 19 |
-
This
|
| 20 |
-
|
| 21 |
-
|
| 22 |
-
|
| 23 |
-
|
| 24 |
-
|
| 25 |
-
|
| 26 |
-
|
| 27 |
-
|
| 28 |
-
|
| 29 |
-
|
| 30 |
-
|
| 31 |
-
|
| 32 |
-
|
| 33 |
-
|
| 34 |
-
|
| 35 |
-
|
| 36 |
-
|
| 37 |
-
|
| 38 |
-
|
|
| 39 |
-
|
| 40 |
-
|
| 41 |
-
|
| 42 |
-
|
| 43 |
-
|
| 44 |
-
|
| 45 |
-
|
| 46 |
-
|
| 47 |
-
|
| 48 |
-
|
| 49 |
-
|
| 50 |
-
|
| 51 |
-
|
| 52 |
-
|
| 53 |
-
|
| 54 |
-
|
| 55 |
-
|
| 56 |
-
|
| 57 |
-
|
| 58 |
-
|
| 59 |
-
|
| 60 |
-
|
| 61 |
-
|
| 62 |
-
|
| 63 |
-
|
| 64 |
-
|
| 65 |
-
|
| 66 |
-
|
|
| 67 |
-
|
|
| 68 |
-
| Inception-v3 |
|
| 69 |
-
| Inception-v3 |
|
| 70 |
-
| Inception-v3 |
|
| 71 |
-
| Inception-v3 |
|
| 72 |
-
| Inception-v3 |
|
| 73 |
-
| Inception-v3 |
|
| 74 |
-
| Inception-v3 |
|
| 75 |
-
| Inception-v3 |
|
| 76 |
-
| Inception-v3 |
|
| 77 |
-
| Inception-v3 |
|
| 78 |
-
| Inception-v3 |
|
| 79 |
-
| Inception-v3 |
|
| 80 |
-
| Inception-v3 |
|
| 81 |
-
| Inception-v3 |
|
| 82 |
-
| Inception-v3 |
|
| 83 |
-
| Inception-v3 |
|
| 84 |
-
| Inception-v3 |
|
| 85 |
-
| Inception-v3 |
|
| 86 |
-
| Inception-v3 |
|
| 87 |
-
| Inception-v3 |
|
| 88 |
-
| Inception-v3 |
|
| 89 |
-
| Inception-v3 |
|
| 90 |
-
| Inception-v3 |
|
| 91 |
-
| Inception-v3 |
|
| 92 |
-
| Inception-v3 |
|
| 93 |
-
| Inception-v3 |
|
| 94 |
-
| Inception-v3 |
|
| 95 |
-
| Inception-v3 |
|
| 96 |
-
| Inception-v3 |
|
| 97 |
-
| Inception-v3 |
|
| 98 |
-
| Inception-v3 |
|
| 99 |
-
| Inception-v3 |
|
| 100 |
-
| Inception-v3 |
|
| 101 |
-
|
| 102 |
-
|
| 103 |
-
|
| 104 |
-
|
| 105 |
-
|
| 106 |
-
|
| 107 |
-
|
| 108 |
-
|
| 109 |
-
|
| 110 |
-
|
| 111 |
-
|
| 112 |
-
|
| 113 |
-
|
| 114 |
-
|
| 115 |
-
|
| 116 |
-
|
| 117 |
-
|
| 118 |
-
|
| 119 |
-
|
| 120 |
-
|
| 121 |
-
|
| 122 |
-
|
| 123 |
-
|
| 124 |
-
|
| 125 |
-
|
| 126 |
-
|
| 127 |
-
|
| 128 |
-
|
| 129 |
-
|
| 130 |
-
|
| 131 |
-
weights and runs this model on a sample input.
|
| 132 |
-
|
| 133 |
-
```bash
|
| 134 |
-
python -m qai_hub_models.models.inception_v3.demo
|
| 135 |
-
```
|
| 136 |
-
|
| 137 |
-
The above demo runs a reference implementation of pre-processing, model
|
| 138 |
-
inference, and post processing.
|
| 139 |
-
|
| 140 |
-
**NOTE**: If you want running in a Jupyter Notebook or Google Colab like
|
| 141 |
-
environment, please add the following to your cell (instead of the above).
|
| 142 |
-
```
|
| 143 |
-
%run -m qai_hub_models.models.inception_v3.demo
|
| 144 |
-
```
|
| 145 |
-
|
| 146 |
-
|
| 147 |
-
### Run model on a cloud-hosted device
|
| 148 |
-
|
| 149 |
-
In addition to the demo, you can also run the model on a cloud-hosted Qualcomm®
|
| 150 |
-
device. This script does the following:
|
| 151 |
-
* Performance check on-device on a cloud-hosted device
|
| 152 |
-
* Downloads compiled assets that can be deployed on-device for Android.
|
| 153 |
-
* Accuracy check between PyTorch and on-device outputs.
|
| 154 |
-
|
| 155 |
-
```bash
|
| 156 |
-
python -m qai_hub_models.models.inception_v3.export
|
| 157 |
-
```
|
| 158 |
-
|
| 159 |
-
|
| 160 |
-
|
| 161 |
-
## How does this work?
|
| 162 |
-
|
| 163 |
-
This [export script](https://aihub.qualcomm.com/models/inception_v3/qai_hub_models/models/Inception-v3/export.py)
|
| 164 |
-
leverages [Qualcomm® AI Hub](https://aihub.qualcomm.com/) to optimize, validate, and deploy this model
|
| 165 |
-
on-device. Lets go through each step below in detail:
|
| 166 |
-
|
| 167 |
-
Step 1: **Compile model for on-device deployment**
|
| 168 |
-
|
| 169 |
-
To compile a PyTorch model for on-device deployment, we first trace the model
|
| 170 |
-
in memory using the `jit.trace` and then call the `submit_compile_job` API.
|
| 171 |
-
|
| 172 |
-
```python
|
| 173 |
-
import torch
|
| 174 |
-
|
| 175 |
-
import qai_hub as hub
|
| 176 |
-
from qai_hub_models.models.inception_v3 import Model
|
| 177 |
-
|
| 178 |
-
# Load the model
|
| 179 |
-
torch_model = Model.from_pretrained()
|
| 180 |
-
|
| 181 |
-
# Device
|
| 182 |
-
device = hub.Device("Samsung Galaxy S25")
|
| 183 |
-
|
| 184 |
-
# Trace model
|
| 185 |
-
input_shape = torch_model.get_input_spec()
|
| 186 |
-
sample_inputs = torch_model.sample_inputs()
|
| 187 |
-
|
| 188 |
-
pt_model = torch.jit.trace(torch_model, [torch.tensor(data[0]) for _, data in sample_inputs.items()])
|
| 189 |
-
|
| 190 |
-
# Compile model on a specific device
|
| 191 |
-
compile_job = hub.submit_compile_job(
|
| 192 |
-
model=pt_model,
|
| 193 |
-
device=device,
|
| 194 |
-
input_specs=torch_model.get_input_spec(),
|
| 195 |
-
)
|
| 196 |
-
|
| 197 |
-
# Get target model to run on-device
|
| 198 |
-
target_model = compile_job.get_target_model()
|
| 199 |
-
|
| 200 |
-
```
|
| 201 |
-
|
| 202 |
-
|
| 203 |
-
Step 2: **Performance profiling on cloud-hosted device**
|
| 204 |
-
|
| 205 |
-
After compiling models from step 1. Models can be profiled model on-device using the
|
| 206 |
-
`target_model`. Note that this scripts runs the model on a device automatically
|
| 207 |
-
provisioned in the cloud. Once the job is submitted, you can navigate to a
|
| 208 |
-
provided job URL to view a variety of on-device performance metrics.
|
| 209 |
-
```python
|
| 210 |
-
profile_job = hub.submit_profile_job(
|
| 211 |
-
model=target_model,
|
| 212 |
-
device=device,
|
| 213 |
-
)
|
| 214 |
-
|
| 215 |
-
```
|
| 216 |
-
|
| 217 |
-
Step 3: **Verify on-device accuracy**
|
| 218 |
-
|
| 219 |
-
To verify the accuracy of the model on-device, you can run on-device inference
|
| 220 |
-
on sample input data on the same cloud hosted device.
|
| 221 |
-
```python
|
| 222 |
-
input_data = torch_model.sample_inputs()
|
| 223 |
-
inference_job = hub.submit_inference_job(
|
| 224 |
-
model=target_model,
|
| 225 |
-
device=device,
|
| 226 |
-
inputs=input_data,
|
| 227 |
-
)
|
| 228 |
-
on_device_output = inference_job.download_output_data()
|
| 229 |
-
|
| 230 |
-
```
|
| 231 |
-
With the output of the model, you can compute like PSNR, relative errors or
|
| 232 |
-
spot check the output with expected output.
|
| 233 |
-
|
| 234 |
-
**Note**: This on-device profiling and inference requires access to Qualcomm®
|
| 235 |
-
AI Hub Workbench. [Sign up for access](https://myaccount.qualcomm.com/signup).
|
| 236 |
-
|
| 237 |
-
|
| 238 |
-
|
| 239 |
-
## Run demo on a cloud-hosted device
|
| 240 |
-
|
| 241 |
-
You can also run the demo on-device.
|
| 242 |
-
|
| 243 |
-
```bash
|
| 244 |
-
python -m qai_hub_models.models.inception_v3.demo --eval-mode on-device
|
| 245 |
-
```
|
| 246 |
-
|
| 247 |
-
**NOTE**: If you want running in a Jupyter Notebook or Google Colab like
|
| 248 |
-
environment, please add the following to your cell (instead of the above).
|
| 249 |
-
```
|
| 250 |
-
%run -m qai_hub_models.models.inception_v3.demo -- --eval-mode on-device
|
| 251 |
-
```
|
| 252 |
-
|
| 253 |
-
|
| 254 |
-
## Deploying compiled model to Android
|
| 255 |
-
|
| 256 |
-
|
| 257 |
-
The models can be deployed using multiple runtimes:
|
| 258 |
-
- TensorFlow Lite (`.tflite` export): [This
|
| 259 |
-
tutorial](https://www.tensorflow.org/lite/android/quickstart) provides a
|
| 260 |
-
guide to deploy the .tflite model in an Android application.
|
| 261 |
-
|
| 262 |
-
|
| 263 |
-
- QNN (`.so` export ): This [sample
|
| 264 |
-
app](https://docs.qualcomm.com/bundle/publicresource/topics/80-63442-50/sample_app.html)
|
| 265 |
-
provides instructions on how to use the `.so` shared library in an Android application.
|
| 266 |
-
|
| 267 |
-
|
| 268 |
-
## View on Qualcomm® AI Hub
|
| 269 |
-
Get more details on Inception-v3's performance across various devices [here](https://aihub.qualcomm.com/models/inception_v3).
|
| 270 |
-
Explore all available models on [Qualcomm® AI Hub](https://aihub.qualcomm.com/)
|
| 271 |
-
|
| 272 |
|
| 273 |
## License
|
| 274 |
* The license for the original implementation of Inception-v3 can be found
|
| 275 |
[here](https://github.com/pytorch/vision/blob/main/LICENSE).
|
| 276 |
|
| 277 |
-
|
| 278 |
-
|
| 279 |
## References
|
| 280 |
* [Rethinking the Inception Architecture for Computer Vision](http://arxiv.org/abs/1512.00567)
|
| 281 |
* [Source Model Implementation](https://github.com/pytorch/vision/blob/main/torchvision/models/inception.py)
|
| 282 |
|
| 283 |
-
|
| 284 |
-
|
| 285 |
## Community
|
| 286 |
* Join [our AI Hub Slack community](https://aihub.qualcomm.com/community/slack) to collaborate, post questions and learn more about on-device AI.
|
| 287 |
* For questions or feedback please [reach out to us](mailto:ai-hub-support@qti.qualcomm.com).
|
| 288 |
-
|
| 289 |
-
|
|
|
|
| 10 |
|
| 11 |
|
| 12 |
|
| 13 |
+
# Inception-v3: Optimized for Qualcomm Devices
|
|
|
|
|
|
|
| 14 |
|
| 15 |
InceptionNetV3 is a machine learning model that can classify images from the Imagenet dataset. It can also be used as a backbone in building more complex models for specific use cases.
|
| 16 |
|
| 17 |
+
This is based on the implementation of Inception-v3 found [here](https://github.com/pytorch/vision/blob/main/torchvision/models/inception.py).
|
| 18 |
+
This repository contains pre-exported model files optimized for Qualcomm® devices. You can use the [Qualcomm® AI Hub Models](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/models/inception_v3) library to export with custom configurations. More details on model performance across various devices, can be found [here](#performance-summary).
|
| 19 |
+
|
| 20 |
+
Qualcomm AI Hub Models uses [Qualcomm AI Hub Workbench](https://workbench.aihub.qualcomm.com) to compile, profile, and evaluate this model. [Sign up](https://myaccount.qualcomm.com/signup) to run these models on a hosted Qualcomm® device.
|
| 21 |
+
|
| 22 |
+
## Getting Started
|
| 23 |
+
There are two ways to deploy this model on your device:
|
| 24 |
+
|
| 25 |
+
### Option 1: Download Pre-Exported Models
|
| 26 |
+
|
| 27 |
+
Below are pre-exported model assets ready for deployment.
|
| 28 |
+
|
| 29 |
+
| Runtime | Precision | Chipset | SDK Versions | Download |
|
| 30 |
+
|---|---|---|---|---|
|
| 31 |
+
| ONNX | float | Universal | QAIRT 2.37, ONNX Runtime 1.23.0 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/inception_v3/releases/v0.46.1/inception_v3-onnx-float.zip)
|
| 32 |
+
| ONNX | w8a8 | Universal | QAIRT 2.37, ONNX Runtime 1.23.0 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/inception_v3/releases/v0.46.1/inception_v3-onnx-w8a8.zip)
|
| 33 |
+
| QNN_DLC | float | Universal | QAIRT 2.42 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/inception_v3/releases/v0.46.1/inception_v3-qnn_dlc-float.zip)
|
| 34 |
+
| QNN_DLC | w8a8 | Universal | QAIRT 2.42 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/inception_v3/releases/v0.46.1/inception_v3-qnn_dlc-w8a8.zip)
|
| 35 |
+
| TFLITE | float | Universal | QAIRT 2.42, TFLite 2.17.0 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/inception_v3/releases/v0.46.1/inception_v3-tflite-float.zip)
|
| 36 |
+
| TFLITE | w8a8 | Universal | QAIRT 2.42, TFLite 2.17.0 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/inception_v3/releases/v0.46.1/inception_v3-tflite-w8a8.zip)
|
| 37 |
+
|
| 38 |
+
For more device-specific assets and performance metrics, visit **[Inception-v3 on Qualcomm® AI Hub](https://aihub.qualcomm.com/models/inception_v3)**.
|
| 39 |
+
|
| 40 |
+
|
| 41 |
+
### Option 2: Export with Custom Configurations
|
| 42 |
+
|
| 43 |
+
Use the [Qualcomm® AI Hub Models](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/models/inception_v3) Python library to compile and export the model with your own:
|
| 44 |
+
- Custom weights (e.g., fine-tuned checkpoints)
|
| 45 |
+
- Custom input shapes
|
| 46 |
+
- Target device and runtime configurations
|
| 47 |
+
|
| 48 |
+
This option is ideal if you need to customize the model beyond the default configuration provided here.
|
| 49 |
+
|
| 50 |
+
See our repository for [Inception-v3 on GitHub](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/models/inception_v3) for usage instructions.
|
| 51 |
+
|
| 52 |
+
## Model Details
|
| 53 |
+
|
| 54 |
+
**Model Type:** Model_use_case.image_classification
|
| 55 |
+
|
| 56 |
+
**Model Stats:**
|
| 57 |
+
- Model checkpoint: Imagenet
|
| 58 |
+
- Input resolution: 224x224
|
| 59 |
+
- Number of parameters: 23.9M
|
| 60 |
+
- Model size (float): 90.9 MB
|
| 61 |
+
- Model size (w8a8): 23.3 MB
|
| 62 |
+
|
| 63 |
+
## Performance Summary
|
| 64 |
+
| Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit
|
| 65 |
+
|---|---|---|---|---|---|---
|
| 66 |
+
| Inception-v3 | ONNX | float | Snapdragon® X Elite | 1.518 ms | 46 - 46 MB | NPU
|
| 67 |
+
| Inception-v3 | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 1.26 ms | 0 - 122 MB | NPU
|
| 68 |
+
| Inception-v3 | ONNX | float | Qualcomm® QCS8550 (Proxy) | 1.66 ms | 0 - 54 MB | NPU
|
| 69 |
+
| Inception-v3 | ONNX | float | Qualcomm® QCS9075 | 2.373 ms | 0 - 4 MB | NPU
|
| 70 |
+
| Inception-v3 | ONNX | float | Snapdragon® 8 Elite For Galaxy Mobile | 1.04 ms | 0 - 102 MB | NPU
|
| 71 |
+
| Inception-v3 | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 0.917 ms | 0 - 103 MB | NPU
|
| 72 |
+
| Inception-v3 | ONNX | w8a8 | Snapdragon® X Elite | 0.702 ms | 24 - 24 MB | NPU
|
| 73 |
+
| Inception-v3 | ONNX | w8a8 | Snapdragon® 8 Gen 3 Mobile | 0.672 ms | 0 - 162 MB | NPU
|
| 74 |
+
| Inception-v3 | ONNX | w8a8 | Qualcomm® QCS6490 | 24.544 ms | 8 - 41 MB | CPU
|
| 75 |
+
| Inception-v3 | ONNX | w8a8 | Qualcomm® QCS8550 (Proxy) | 0.86 ms | 0 - 225 MB | NPU
|
| 76 |
+
| Inception-v3 | ONNX | w8a8 | Qualcomm® QCS9075 | 0.947 ms | 0 - 3 MB | NPU
|
| 77 |
+
| Inception-v3 | ONNX | w8a8 | Qualcomm® QCM6690 | 17.95 ms | 10 - 20 MB | CPU
|
| 78 |
+
| Inception-v3 | ONNX | w8a8 | Snapdragon® 8 Elite For Galaxy Mobile | 0.557 ms | 0 - 123 MB | NPU
|
| 79 |
+
| Inception-v3 | ONNX | w8a8 | Snapdragon® 7 Gen 4 Mobile | 13.886 ms | 10 - 20 MB | CPU
|
| 80 |
+
| Inception-v3 | ONNX | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 0.521 ms | 0 - 124 MB | NPU
|
| 81 |
+
| Inception-v3 | QNN_DLC | float | Snapdragon® X Elite | 1.426 ms | 1 - 1 MB | NPU
|
| 82 |
+
| Inception-v3 | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 1.02 ms | 0 - 56 MB | NPU
|
| 83 |
+
| Inception-v3 | QNN_DLC | float | Qualcomm® QCS8275 (Proxy) | 7.735 ms | 1 - 32 MB | NPU
|
| 84 |
+
| Inception-v3 | QNN_DLC | float | Qualcomm® QCS8550 (Proxy) | 1.374 ms | 1 - 192 MB | NPU
|
| 85 |
+
| Inception-v3 | QNN_DLC | float | Qualcomm® SA8775P | 2.203 ms | 0 - 35 MB | NPU
|
| 86 |
+
| Inception-v3 | QNN_DLC | float | Qualcomm® QCS9075 | 2.048 ms | 1 - 3 MB | NPU
|
| 87 |
+
| Inception-v3 | QNN_DLC | float | Qualcomm® QCS8450 (Proxy) | 2.342 ms | 0 - 55 MB | NPU
|
| 88 |
+
| Inception-v3 | QNN_DLC | float | Qualcomm® SA7255P | 7.735 ms | 1 - 32 MB | NPU
|
| 89 |
+
| Inception-v3 | QNN_DLC | float | Qualcomm® SA8295P | 2.628 ms | 0 - 31 MB | NPU
|
| 90 |
+
| Inception-v3 | QNN_DLC | float | Snapdragon® 8 Elite For Galaxy Mobile | 0.849 ms | 1 - 35 MB | NPU
|
| 91 |
+
| Inception-v3 | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 Mobile | 0.764 ms | 1 - 37 MB | NPU
|
| 92 |
+
| Inception-v3 | QNN_DLC | w8a8 | Snapdragon® X Elite | 0.693 ms | 0 - 0 MB | NPU
|
| 93 |
+
| Inception-v3 | QNN_DLC | w8a8 | Snapdragon® 8 Gen 3 Mobile | 0.48 ms | 0 - 77 MB | NPU
|
| 94 |
+
| Inception-v3 | QNN_DLC | w8a8 | Qualcomm® QCS6490 | 2.29 ms | 0 - 2 MB | NPU
|
| 95 |
+
| Inception-v3 | QNN_DLC | w8a8 | Qualcomm® QCS8275 (Proxy) | 1.527 ms | 0 - 47 MB | NPU
|
| 96 |
+
| Inception-v3 | QNN_DLC | w8a8 | Qualcomm® QCS8550 (Proxy) | 0.629 ms | 0 - 2 MB | NPU
|
| 97 |
+
| Inception-v3 | QNN_DLC | w8a8 | Qualcomm® SA8775P | 0.779 ms | 0 - 51 MB | NPU
|
| 98 |
+
| Inception-v3 | QNN_DLC | w8a8 | Qualcomm® QCS9075 | 0.62 ms | 2 - 4 MB | NPU
|
| 99 |
+
| Inception-v3 | QNN_DLC | w8a8 | Qualcomm® QCM6690 | 5.798 ms | 0 - 50 MB | NPU
|
| 100 |
+
| Inception-v3 | QNN_DLC | w8a8 | Qualcomm® QCS8450 (Proxy) | 0.896 ms | 0 - 74 MB | NPU
|
| 101 |
+
| Inception-v3 | QNN_DLC | w8a8 | Qualcomm® SA7255P | 1.527 ms | 0 - 47 MB | NPU
|
| 102 |
+
| Inception-v3 | QNN_DLC | w8a8 | Qualcomm® SA8295P | 1.134 ms | 0 - 45 MB | NPU
|
| 103 |
+
| Inception-v3 | QNN_DLC | w8a8 | Snapdragon® 8 Elite For Galaxy Mobile | 0.404 ms | 0 - 47 MB | NPU
|
| 104 |
+
| Inception-v3 | QNN_DLC | w8a8 | Snapdragon® 7 Gen 4 Mobile | 0.983 ms | 0 - 50 MB | NPU
|
| 105 |
+
| Inception-v3 | QNN_DLC | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 0.368 ms | 0 - 51 MB | NPU
|
| 106 |
+
| Inception-v3 | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 1.026 ms | 0 - 125 MB | NPU
|
| 107 |
+
| Inception-v3 | TFLITE | float | Qualcomm® QCS8275 (Proxy) | 7.782 ms | 0 - 73 MB | NPU
|
| 108 |
+
| Inception-v3 | TFLITE | float | Qualcomm® QCS8550 (Proxy) | 1.375 ms | 0 - 3 MB | NPU
|
| 109 |
+
| Inception-v3 | TFLITE | float | Qualcomm® SA8775P | 2.238 ms | 0 - 75 MB | NPU
|
| 110 |
+
| Inception-v3 | TFLITE | float | Qualcomm® QCS9075 | 2.047 ms | 0 - 48 MB | NPU
|
| 111 |
+
| Inception-v3 | TFLITE | float | Qualcomm® QCS8450 (Proxy) | 2.344 ms | 0 - 126 MB | NPU
|
| 112 |
+
| Inception-v3 | TFLITE | float | Qualcomm® SA7255P | 7.782 ms | 0 - 73 MB | NPU
|
| 113 |
+
| Inception-v3 | TFLITE | float | Qualcomm® SA8295P | 2.64 ms | 0 - 69 MB | NPU
|
| 114 |
+
| Inception-v3 | TFLITE | float | Snapdragon® 8 Elite For Galaxy Mobile | 0.871 ms | 0 - 77 MB | NPU
|
| 115 |
+
| Inception-v3 | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 0.772 ms | 0 - 77 MB | NPU
|
| 116 |
+
| Inception-v3 | TFLITE | w8a8 | Snapdragon® 8 Gen 3 Mobile | 0.484 ms | 0 - 73 MB | NPU
|
| 117 |
+
| Inception-v3 | TFLITE | w8a8 | Qualcomm® QCS6490 | 2.265 ms | 0 - 25 MB | NPU
|
| 118 |
+
| Inception-v3 | TFLITE | w8a8 | Qualcomm® QCS8275 (Proxy) | 1.616 ms | 0 - 47 MB | NPU
|
| 119 |
+
| Inception-v3 | TFLITE | w8a8 | Qualcomm® QCS8550 (Proxy) | 0.644 ms | 0 - 2 MB | NPU
|
| 120 |
+
| Inception-v3 | TFLITE | w8a8 | Qualcomm® SA8775P | 0.852 ms | 0 - 50 MB | NPU
|
| 121 |
+
| Inception-v3 | TFLITE | w8a8 | Qualcomm® QCS9075 | 0.699 ms | 0 - 25 MB | NPU
|
| 122 |
+
| Inception-v3 | TFLITE | w8a8 | Qualcomm® QCM6690 | 5.787 ms | 0 - 49 MB | NPU
|
| 123 |
+
| Inception-v3 | TFLITE | w8a8 | Qualcomm® QCS8450 (Proxy) | 0.945 ms | 0 - 75 MB | NPU
|
| 124 |
+
| Inception-v3 | TFLITE | w8a8 | Qualcomm® SA7255P | 1.616 ms | 0 - 47 MB | NPU
|
| 125 |
+
| Inception-v3 | TFLITE | w8a8 | Qualcomm® SA8295P | 1.251 ms | 0 - 45 MB | NPU
|
| 126 |
+
| Inception-v3 | TFLITE | w8a8 | Snapdragon® 8 Elite For Galaxy Mobile | 0.432 ms | 0 - 50 MB | NPU
|
| 127 |
+
| Inception-v3 | TFLITE | w8a8 | Snapdragon® 7 Gen 4 Mobile | 1.007 ms | 0 - 48 MB | NPU
|
| 128 |
+
| Inception-v3 | TFLITE | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 0.379 ms | 0 - 50 MB | NPU
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 129 |
|
| 130 |
## License
|
| 131 |
* The license for the original implementation of Inception-v3 can be found
|
| 132 |
[here](https://github.com/pytorch/vision/blob/main/LICENSE).
|
| 133 |
|
|
|
|
|
|
|
| 134 |
## References
|
| 135 |
* [Rethinking the Inception Architecture for Computer Vision](http://arxiv.org/abs/1512.00567)
|
| 136 |
* [Source Model Implementation](https://github.com/pytorch/vision/blob/main/torchvision/models/inception.py)
|
| 137 |
|
|
|
|
|
|
|
| 138 |
## Community
|
| 139 |
* Join [our AI Hub Slack community](https://aihub.qualcomm.com/community/slack) to collaborate, post questions and learn more about on-device AI.
|
| 140 |
* For questions or feedback please [reach out to us](mailto:ai-hub-support@qti.qualcomm.com).
|
|
|
|
|
|
tool-versions.yaml
DELETED
|
@@ -1,4 +0,0 @@
|
|
| 1 |
-
tool_versions:
|
| 2 |
-
onnx:
|
| 3 |
-
qairt: 2.37.1.250807093845_124904
|
| 4 |
-
onnx_runtime: 1.23.0
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|