v0.30.5
Browse filesSee https://github.com/quic/ai-hub-models/releases/v0.30.5 for changelog.
README.md
CHANGED
|
@@ -34,23 +34,23 @@ More details on model performance across various devices, can be found
|
|
| 34 |
|
| 35 |
| Model | Precision | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit | Target Model
|
| 36 |
|---|---|---|---|---|---|---|---|---|
|
| 37 |
-
| DETR-ResNet50-DC5 | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN |
|
| 38 |
-
| DETR-ResNet50-DC5 | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN |
|
| 39 |
-
| DETR-ResNet50-DC5 | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN |
|
| 40 |
-
| DETR-ResNet50-DC5 | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN |
|
| 41 |
-
| DETR-ResNet50-DC5 | float | SA7255P ADP | Qualcomm® SA7255P | QNN |
|
| 42 |
-
| DETR-ResNet50-DC5 | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN |
|
| 43 |
-
| DETR-ResNet50-DC5 | float | SA8295P ADP | Qualcomm® SA8295P | QNN |
|
| 44 |
-
| DETR-ResNet50-DC5 | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN |
|
| 45 |
-
| DETR-ResNet50-DC5 | float | SA8775P ADP | Qualcomm® SA8775P | QNN |
|
| 46 |
-
| DETR-ResNet50-DC5 | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | QNN |
|
| 47 |
-
| DETR-ResNet50-DC5 | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | ONNX |
|
| 48 |
-
| DETR-ResNet50-DC5 | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN |
|
| 49 |
-
| DETR-ResNet50-DC5 | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX |
|
| 50 |
-
| DETR-ResNet50-DC5 | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | QNN |
|
| 51 |
-
| DETR-ResNet50-DC5 | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 33.
|
| 52 |
-
| DETR-ResNet50-DC5 | float | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN |
|
| 53 |
-
| DETR-ResNet50-DC5 | float | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX |
|
| 54 |
|
| 55 |
|
| 56 |
|
|
@@ -114,8 +114,8 @@ Profiling Results
|
|
| 114 |
DETR-ResNet50-DC5
|
| 115 |
Device : cs_8275 (ANDROID 14)
|
| 116 |
Runtime : QNN
|
| 117 |
-
Estimated inference time (ms) :
|
| 118 |
-
Estimated peak memory usage (MB): [
|
| 119 |
Total # Ops : 829
|
| 120 |
Compute Unit(s) : npu (829 ops) gpu (0 ops) cpu (0 ops)
|
| 121 |
```
|
|
@@ -204,13 +204,13 @@ AI Hub. [Sign up for access](https://myaccount.qualcomm.com/signup).
|
|
| 204 |
You can also run the demo on-device.
|
| 205 |
|
| 206 |
```bash
|
| 207 |
-
python -m qai_hub_models.models.detr_resnet50_dc5.demo --on-device
|
| 208 |
```
|
| 209 |
|
| 210 |
**NOTE**: If you want running in a Jupyter Notebook or Google Colab like
|
| 211 |
environment, please add the following to your cell (instead of the above).
|
| 212 |
```
|
| 213 |
-
%run -m qai_hub_models.models.detr_resnet50_dc5.demo -- --on-device
|
| 214 |
```
|
| 215 |
|
| 216 |
|
|
|
|
| 34 |
|
| 35 |
| Model | Precision | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit | Target Model
|
| 36 |
|---|---|---|---|---|---|---|---|---|
|
| 37 |
+
| DETR-ResNet50-DC5 | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN | 1051.971 ms | 0 - 9 MB | NPU | Use Export Script |
|
| 38 |
+
| DETR-ResNet50-DC5 | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN | 76.458 ms | 5 - 135 MB | NPU | Use Export Script |
|
| 39 |
+
| DETR-ResNet50-DC5 | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN | 47.113 ms | 5 - 7 MB | NPU | Use Export Script |
|
| 40 |
+
| DETR-ResNet50-DC5 | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN | 59.899 ms | 1 - 11 MB | NPU | Use Export Script |
|
| 41 |
+
| DETR-ResNet50-DC5 | float | SA7255P ADP | Qualcomm® SA7255P | QNN | 1051.971 ms | 0 - 9 MB | NPU | Use Export Script |
|
| 42 |
+
| DETR-ResNet50-DC5 | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN | 47.139 ms | 5 - 7 MB | NPU | Use Export Script |
|
| 43 |
+
| DETR-ResNet50-DC5 | float | SA8295P ADP | Qualcomm® SA8295P | QNN | 66.384 ms | 0 - 17 MB | NPU | Use Export Script |
|
| 44 |
+
| DETR-ResNet50-DC5 | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN | 47.145 ms | 5 - 7 MB | NPU | Use Export Script |
|
| 45 |
+
| DETR-ResNet50-DC5 | float | SA8775P ADP | Qualcomm® SA8775P | QNN | 59.899 ms | 1 - 11 MB | NPU | Use Export Script |
|
| 46 |
+
| DETR-ResNet50-DC5 | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | QNN | 47.548 ms | 5 - 61 MB | NPU | Use Export Script |
|
| 47 |
+
| DETR-ResNet50-DC5 | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | ONNX | 50.299 ms | 0 - 264 MB | NPU | [DETR-ResNet50-DC5.onnx](https://huggingface.co/qualcomm/DETR-ResNet50-DC5/blob/main/DETR-ResNet50-DC5.onnx) |
|
| 48 |
+
| DETR-ResNet50-DC5 | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN | 34.925 ms | 5 - 263 MB | NPU | Use Export Script |
|
| 49 |
+
| DETR-ResNet50-DC5 | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 37.821 ms | 5 - 246 MB | NPU | [DETR-ResNet50-DC5.onnx](https://huggingface.co/qualcomm/DETR-ResNet50-DC5/blob/main/DETR-ResNet50-DC5.onnx) |
|
| 50 |
+
| DETR-ResNet50-DC5 | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | QNN | 30.777 ms | 5 - 272 MB | NPU | Use Export Script |
|
| 51 |
+
| DETR-ResNet50-DC5 | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 33.129 ms | 11 - 183 MB | NPU | [DETR-ResNet50-DC5.onnx](https://huggingface.co/qualcomm/DETR-ResNet50-DC5/blob/main/DETR-ResNet50-DC5.onnx) |
|
| 52 |
+
| DETR-ResNet50-DC5 | float | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN | 47.516 ms | 5 - 5 MB | NPU | Use Export Script |
|
| 53 |
+
| DETR-ResNet50-DC5 | float | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 52.356 ms | 80 - 80 MB | NPU | [DETR-ResNet50-DC5.onnx](https://huggingface.co/qualcomm/DETR-ResNet50-DC5/blob/main/DETR-ResNet50-DC5.onnx) |
|
| 54 |
|
| 55 |
|
| 56 |
|
|
|
|
| 114 |
DETR-ResNet50-DC5
|
| 115 |
Device : cs_8275 (ANDROID 14)
|
| 116 |
Runtime : QNN
|
| 117 |
+
Estimated inference time (ms) : 1052.0
|
| 118 |
+
Estimated peak memory usage (MB): [0, 9]
|
| 119 |
Total # Ops : 829
|
| 120 |
Compute Unit(s) : npu (829 ops) gpu (0 ops) cpu (0 ops)
|
| 121 |
```
|
|
|
|
| 204 |
You can also run the demo on-device.
|
| 205 |
|
| 206 |
```bash
|
| 207 |
+
python -m qai_hub_models.models.detr_resnet50_dc5.demo --eval-mode on-device
|
| 208 |
```
|
| 209 |
|
| 210 |
**NOTE**: If you want running in a Jupyter Notebook or Google Colab like
|
| 211 |
environment, please add the following to your cell (instead of the above).
|
| 212 |
```
|
| 213 |
+
%run -m qai_hub_models.models.detr_resnet50_dc5.demo -- --eval-mode on-device
|
| 214 |
```
|
| 215 |
|
| 216 |
|