qaihm-bot commited on
Commit
07f3168
·
verified ·
1 Parent(s): f59f478

See https://github.com/qualcomm/ai-hub-models/releases/v0.49.1 for changelog.

Files changed (1) hide show
  1. README.md +35 -35
README.md CHANGED
@@ -15,7 +15,7 @@ pipeline_tag: object-detection
15
  DETR is a machine learning model that can detect objects (trained on COCO dataset).
16
 
17
  This is based on the implementation of DETR-ResNet50 found [here](https://github.com/facebookresearch/detr).
18
- This repository contains pre-exported model files optimized for Qualcomm® devices. You can use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/main/qai_hub_models/models/detr_resnet50) 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
 
@@ -28,23 +28,23 @@ Below are pre-exported model assets ready for deployment.
28
 
29
  | Runtime | Precision | Chipset | SDK Versions | Download |
30
  |---|---|---|---|---|
31
- | ONNX | float | Universal | QAIRT 2.42, ONNX Runtime 1.24.1 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/detr_resnet50/releases/v0.48.0/detr_resnet50-onnx-float.zip)
32
- | QNN_DLC | float | Universal | QAIRT 2.43 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/detr_resnet50/releases/v0.48.0/detr_resnet50-qnn_dlc-float.zip)
33
- | TFLITE | float | Universal | QAIRT 2.43, TFLite 2.17.0 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/detr_resnet50/releases/v0.48.0/detr_resnet50-tflite-float.zip)
34
 
35
  For more device-specific assets and performance metrics, visit **[DETR-ResNet50 on Qualcomm® AI Hub](https://aihub.qualcomm.com/models/detr_resnet50)**.
36
 
37
 
38
  ### Option 2: Export with Custom Configurations
39
 
40
- Use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/main/qai_hub_models/models/detr_resnet50) Python library to compile and export the model with your own:
41
  - Custom weights (e.g., fine-tuned checkpoints)
42
  - Custom input shapes
43
  - Target device and runtime configurations
44
 
45
  This option is ideal if you need to customize the model beyond the default configuration provided here.
46
 
47
- See our repository for [DETR-ResNet50 on GitHub](https://github.com/qualcomm/ai-hub-models/blob/main/qai_hub_models/models/detr_resnet50) for usage instructions.
48
 
49
  ## Model Details
50
 
@@ -59,35 +59,35 @@ See our repository for [DETR-ResNet50 on GitHub](https://github.com/qualcomm/ai-
59
  ## Performance Summary
60
  | Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit
61
  |---|---|---|---|---|---|---
62
- | DETR-ResNet50 | ONNX | float | Snapdragon® X2 Elite | 8.368 ms | 77 - 77 MB | NPU
63
- | DETR-ResNet50 | ONNX | float | Snapdragon® X Elite | 18.005 ms | 77 - 77 MB | NPU
64
- | DETR-ResNet50 | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 13.518 ms | 1 - 397 MB | NPU
65
- | DETR-ResNet50 | ONNX | float | Qualcomm® QCS8550 (Proxy) | 17.884 ms | 0 - 95 MB | NPU
66
- | DETR-ResNet50 | ONNX | float | Qualcomm® QCS9075 | 28.427 ms | 5 - 12 MB | NPU
67
- | DETR-ResNet50 | ONNX | float | Snapdragon® 8 Elite For Galaxy Mobile | 9.864 ms | 2 - 316 MB | NPU
68
- | DETR-ResNet50 | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 7.452 ms | 0 - 322 MB | NPU
69
- | DETR-ResNet50 | QNN_DLC | float | Snapdragon® X2 Elite | 9.019 ms | 5 - 5 MB | NPU
70
- | DETR-ResNet50 | QNN_DLC | float | Snapdragon® X Elite | 20.668 ms | 5 - 5 MB | NPU
71
- | DETR-ResNet50 | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 14.943 ms | 5 - 388 MB | NPU
72
- | DETR-ResNet50 | QNN_DLC | float | Qualcomm® QCS8275 (Proxy) | 91.478 ms | 1 - 285 MB | NPU
73
- | DETR-ResNet50 | QNN_DLC | float | Qualcomm® QCS8550 (Proxy) | 20.564 ms | 5 - 6 MB | NPU
74
- | DETR-ResNet50 | QNN_DLC | float | Qualcomm® SA8775P | 133.562 ms | 1 - 284 MB | NPU
75
- | DETR-ResNet50 | QNN_DLC | float | Qualcomm® QCS9075 | 30.754 ms | 5 - 11 MB | NPU
76
- | DETR-ResNet50 | QNN_DLC | float | Qualcomm® QCS8450 (Proxy) | 43.242 ms | 2 - 327 MB | NPU
77
- | DETR-ResNet50 | QNN_DLC | float | Qualcomm® SA7255P | 91.478 ms | 1 - 285 MB | NPU
78
- | DETR-ResNet50 | QNN_DLC | float | Qualcomm® SA8295P | 31.396 ms | 0 - 240 MB | NPU
79
- | DETR-ResNet50 | QNN_DLC | float | Snapdragon® 8 Elite For Galaxy Mobile | 10.284 ms | 0 - 294 MB | NPU
80
- | DETR-ResNet50 | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 Mobile | 8.038 ms | 5 - 303 MB | NPU
81
- | DETR-ResNet50 | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 13.895 ms | 0 - 422 MB | NPU
82
- | DETR-ResNet50 | TFLITE | float | Qualcomm® QCS8275 (Proxy) | 85.59 ms | 0 - 323 MB | NPU
83
- | DETR-ResNet50 | TFLITE | float | Qualcomm® QCS8550 (Proxy) | 19.485 ms | 0 - 3 MB | NPU
84
- | DETR-ResNet50 | TFLITE | float | Qualcomm® SA8775P | 26.994 ms | 0 - 382 MB | NPU
85
- | DETR-ResNet50 | TFLITE | float | Qualcomm® QCS9075 | 30.505 ms | 0 - 88 MB | NPU
86
- | DETR-ResNet50 | TFLITE | float | Qualcomm® QCS8450 (Proxy) | 44.292 ms | 0 - 356 MB | NPU
87
- | DETR-ResNet50 | TFLITE | float | Qualcomm® SA7255P | 85.59 ms | 0 - 323 MB | NPU
88
- | DETR-ResNet50 | TFLITE | float | Qualcomm® SA8295P | 31.076 ms | 0 - 270 MB | NPU
89
- | DETR-ResNet50 | TFLITE | float | Snapdragon® 8 Elite For Galaxy Mobile | 10.275 ms | 0 - 337 MB | NPU
90
- | DETR-ResNet50 | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 7.997 ms | 0 - 342 MB | NPU
91
 
92
  ## License
93
  * The license for the original implementation of DETR-ResNet50 can be found
 
15
  DETR is a machine learning model that can detect objects (trained on COCO dataset).
16
 
17
  This is based on the implementation of DETR-ResNet50 found [here](https://github.com/facebookresearch/detr).
18
+ This repository contains pre-exported model files optimized for Qualcomm® devices. You can use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/tree/v0.49.1/qai_hub_models/models/detr_resnet50) 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
 
 
28
 
29
  | Runtime | Precision | Chipset | SDK Versions | Download |
30
  |---|---|---|---|---|
31
+ | ONNX | float | Universal | QAIRT 2.42, ONNX Runtime 1.24.1 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/detr_resnet50/releases/v0.49.1/detr_resnet50-onnx-float.zip)
32
+ | QNN_DLC | float | Universal | QAIRT 2.43 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/detr_resnet50/releases/v0.49.1/detr_resnet50-qnn_dlc-float.zip)
33
+ | TFLITE | float | Universal | QAIRT 2.43, TFLite 2.17.0 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/detr_resnet50/releases/v0.49.1/detr_resnet50-tflite-float.zip)
34
 
35
  For more device-specific assets and performance metrics, visit **[DETR-ResNet50 on Qualcomm® AI Hub](https://aihub.qualcomm.com/models/detr_resnet50)**.
36
 
37
 
38
  ### Option 2: Export with Custom Configurations
39
 
40
+ Use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/tree/v0.49.1/qai_hub_models/models/detr_resnet50) Python library to compile and export the model with your own:
41
  - Custom weights (e.g., fine-tuned checkpoints)
42
  - Custom input shapes
43
  - Target device and runtime configurations
44
 
45
  This option is ideal if you need to customize the model beyond the default configuration provided here.
46
 
47
+ See our repository for [DETR-ResNet50 on GitHub](https://github.com/qualcomm/ai-hub-models/tree/v0.49.1/qai_hub_models/models/detr_resnet50) for usage instructions.
48
 
49
  ## Model Details
50
 
 
59
  ## Performance Summary
60
  | Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit
61
  |---|---|---|---|---|---|---
62
+ | DETR-ResNet50 | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 7.465 ms | 0 - 322 MB | NPU
63
+ | DETR-ResNet50 | ONNX | float | Snapdragon® X2 Elite | 8.41 ms | 77 - 77 MB | NPU
64
+ | DETR-ResNet50 | ONNX | float | Snapdragon® X Elite | 17.963 ms | 77 - 77 MB | NPU
65
+ | DETR-ResNet50 | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 13.419 ms | 0 - 391 MB | NPU
66
+ | DETR-ResNet50 | ONNX | float | Qualcomm® QCS8550 (Proxy) | 17.844 ms | 0 - 94 MB | NPU
67
+ | DETR-ResNet50 | ONNX | float | Qualcomm® QCS9075 | 28.273 ms | 5 - 12 MB | NPU
68
+ | DETR-ResNet50 | ONNX | float | Snapdragon® 8 Elite For Galaxy Mobile | 9.899 ms | 2 - 311 MB | NPU
69
+ | DETR-ResNet50 | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 Mobile | 8.042 ms | 5 - 303 MB | NPU
70
+ | DETR-ResNet50 | QNN_DLC | float | Snapdragon® X2 Elite | 9.035 ms | 5 - 5 MB | NPU
71
+ | DETR-ResNet50 | QNN_DLC | float | Snapdragon® X Elite | 20.604 ms | 5 - 5 MB | NPU
72
+ | DETR-ResNet50 | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 14.926 ms | 3 - 386 MB | NPU
73
+ | DETR-ResNet50 | QNN_DLC | float | Qualcomm® QCS8275 (Proxy) | 91.542 ms | 2 - 285 MB | NPU
74
+ | DETR-ResNet50 | QNN_DLC | float | Qualcomm® QCS8550 (Proxy) | 20.523 ms | 5 - 7 MB | NPU
75
+ | DETR-ResNet50 | QNN_DLC | float | Qualcomm® SA8775P | 29.405 ms | 1 - 285 MB | NPU
76
+ | DETR-ResNet50 | QNN_DLC | float | Qualcomm® QCS9075 | 30.514 ms | 5 - 11 MB | NPU
77
+ | DETR-ResNet50 | QNN_DLC | float | Qualcomm® QCS8450 (Proxy) | 43.97 ms | 4 - 330 MB | NPU
78
+ | DETR-ResNet50 | QNN_DLC | float | Qualcomm® SA7255P | 91.542 ms | 2 - 285 MB | NPU
79
+ | DETR-ResNet50 | QNN_DLC | float | Qualcomm® SA8295P | 31.387 ms | 0 - 241 MB | NPU
80
+ | DETR-ResNet50 | QNN_DLC | float | Snapdragon® 8 Elite For Galaxy Mobile | 10.303 ms | 5 - 297 MB | NPU
81
+ | DETR-ResNet50 | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 8.021 ms | 0 - 342 MB | NPU
82
+ | DETR-ResNet50 | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 13.89 ms | 0 - 423 MB | NPU
83
+ | DETR-ResNet50 | TFLITE | float | Qualcomm® QCS8275 (Proxy) | 85.562 ms | 0 - 322 MB | NPU
84
+ | DETR-ResNet50 | TFLITE | float | Qualcomm® QCS8550 (Proxy) | 19.636 ms | 0 - 3 MB | NPU
85
+ | DETR-ResNet50 | TFLITE | float | Qualcomm® SA8775P | 26.999 ms | 0 - 382 MB | NPU
86
+ | DETR-ResNet50 | TFLITE | float | Qualcomm® QCS9075 | 30.51 ms | 0 - 88 MB | NPU
87
+ | DETR-ResNet50 | TFLITE | float | Qualcomm® QCS8450 (Proxy) | 44.104 ms | 0 - 357 MB | NPU
88
+ | DETR-ResNet50 | TFLITE | float | Qualcomm® SA7255P | 85.562 ms | 0 - 322 MB | NPU
89
+ | DETR-ResNet50 | TFLITE | float | Qualcomm® SA8295P | 31.083 ms | 0 - 271 MB | NPU
90
+ | DETR-ResNet50 | TFLITE | float | Snapdragon® 8 Elite For Galaxy Mobile | 10.283 ms | 0 - 338 MB | NPU
91
 
92
  ## License
93
  * The license for the original implementation of DETR-ResNet50 can be found