Upload README.md with huggingface_hub
Browse files
README.md
CHANGED
|
@@ -32,10 +32,12 @@ More details on model performance across various devices, can be found
|
|
| 32 |
- Model size: 158 MB
|
| 33 |
|
| 34 |
|
|
|
|
|
|
|
| 35 |
| Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Precision | Primary Compute Unit | Target Model
|
| 36 |
| ---|---|---|---|---|---|---|---|
|
| 37 |
-
| Samsung Galaxy S23 Ultra (Android 13) | Snapdragon® 8 Gen 2 | TFLite |
|
| 38 |
-
|
| 39 |
|
| 40 |
|
| 41 |
## Installation
|
|
@@ -93,19 +95,11 @@ device. This script does the following:
|
|
| 93 |
python -m qai_hub_models.models.detr_resnet50.export
|
| 94 |
```
|
| 95 |
|
| 96 |
-
```
|
| 97 |
-
Profile Job summary of DETR-ResNet50
|
| 98 |
-
--------------------------------------------------
|
| 99 |
-
Device: Snapdragon X Elite CRD (11)
|
| 100 |
-
Estimated Inference Time: 22.40 ms
|
| 101 |
-
Estimated Peak Memory Range: 2.64-2.64 MB
|
| 102 |
-
Compute Units: NPU (863) | Total (863)
|
| 103 |
|
| 104 |
|
| 105 |
-
```
|
| 106 |
## How does this work?
|
| 107 |
|
| 108 |
-
This [export script](https://
|
| 109 |
leverages [Qualcomm® AI Hub](https://aihub.qualcomm.com/) to optimize, validate, and deploy this model
|
| 110 |
on-device. Lets go through each step below in detail:
|
| 111 |
|
|
@@ -182,6 +176,7 @@ spot check the output with expected output.
|
|
| 182 |
AI Hub. [Sign up for access](https://myaccount.qualcomm.com/signup).
|
| 183 |
|
| 184 |
|
|
|
|
| 185 |
## Run demo on a cloud-hosted device
|
| 186 |
|
| 187 |
You can also run the demo on-device.
|
|
@@ -218,7 +213,7 @@ Explore all available models on [Qualcomm® AI Hub](https://aihub.qualcomm.com/)
|
|
| 218 |
## License
|
| 219 |
- The license for the original implementation of DETR-ResNet50 can be found
|
| 220 |
[here](https://github.com/facebookresearch/detr/blob/main/LICENSE).
|
| 221 |
-
- The license for the compiled assets for on-device deployment can be found [here](
|
| 222 |
|
| 223 |
## References
|
| 224 |
* [End-to-End Object Detection with Transformers](https://arxiv.org/abs/2005.12872)
|
|
|
|
| 32 |
- Model size: 158 MB
|
| 33 |
|
| 34 |
|
| 35 |
+
|
| 36 |
+
|
| 37 |
| Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Precision | Primary Compute Unit | Target Model
|
| 38 |
| ---|---|---|---|---|---|---|---|
|
| 39 |
+
| Samsung Galaxy S23 Ultra (Android 13) | Snapdragon® 8 Gen 2 | TFLite | 21.615 ms | 2 - 5 MB | FP16 | NPU | [DETR-ResNet50.tflite](https://huggingface.co/qualcomm/DETR-ResNet50/blob/main/DETR-ResNet50.tflite)
|
| 40 |
+
|
| 41 |
|
| 42 |
|
| 43 |
## Installation
|
|
|
|
| 95 |
python -m qai_hub_models.models.detr_resnet50.export
|
| 96 |
```
|
| 97 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 98 |
|
| 99 |
|
|
|
|
| 100 |
## How does this work?
|
| 101 |
|
| 102 |
+
This [export script](https://aihub.qualcomm.com/models/detr_resnet50/qai_hub_models/models/DETR-ResNet50/export.py)
|
| 103 |
leverages [Qualcomm® AI Hub](https://aihub.qualcomm.com/) to optimize, validate, and deploy this model
|
| 104 |
on-device. Lets go through each step below in detail:
|
| 105 |
|
|
|
|
| 176 |
AI Hub. [Sign up for access](https://myaccount.qualcomm.com/signup).
|
| 177 |
|
| 178 |
|
| 179 |
+
|
| 180 |
## Run demo on a cloud-hosted device
|
| 181 |
|
| 182 |
You can also run the demo on-device.
|
|
|
|
| 213 |
## License
|
| 214 |
- The license for the original implementation of DETR-ResNet50 can be found
|
| 215 |
[here](https://github.com/facebookresearch/detr/blob/main/LICENSE).
|
| 216 |
+
- The license for the compiled assets for on-device deployment can be found [here](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/Qualcomm+AI+Hub+Proprietary+License.pdf)
|
| 217 |
|
| 218 |
## References
|
| 219 |
* [End-to-End Object Detection with Transformers](https://arxiv.org/abs/2005.12872)
|