Hugging Face's logo

qualcomm
/
DETR-ResNet50-DC5

Object Detection
PyTorch
bu_auto
android
Model card Files
xet
Community
qaihm-bot commited on
Commit
fd68b31
·
verified ·
1 Parent(s): c0cdad5

v0.48.0

Browse files

See https://github.com/qualcomm/ai-hub-models/releases/v0.48.0 for changelog.

Files changed (1) hide show
  1. README.md +35 -25
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-DC5 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/quic/ai-hub-models/blob/main/qai_hub_models/models/detr_resnet50_dc5) 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_dc5/releases/v0.47.0/detr_resnet50_dc5-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_dc5/releases/v0.47.0/detr_resnet50_dc5-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_dc5/releases/v0.47.0/detr_resnet50_dc5-tflite-float.zip)
34
 
35
  For more device-specific assets and performance metrics, visit **[DETR-ResNet50-DC5 on Qualcomm® AI Hub](https://aihub.qualcomm.com/models/detr_resnet50_dc5)**.
36
 
37
 
38
  ### Option 2: Export with Custom Configurations
39
 
40
- Use the [Qualcomm® AI Hub Models](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/models/detr_resnet50_dc5) 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-DC5 on GitHub](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/models/detr_resnet50_dc5) for usage instructions.
48
 
49
  ## Model Details
50
 
@@ -58,25 +58,35 @@ See our repository for [DETR-ResNet50-DC5 on GitHub](https://github.com/quic/ai-
58
  ## Performance Summary
59
  | Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit
60
  |---|---|---|---|---|---|---
61
- | DETR-ResNet50-DC5 | ONNX | float | Snapdragon® X Elite | 45.343 ms | 78 - 78 MB | NPU
62
- | DETR-ResNet50-DC5 | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 32.68 ms | 1 - 652 MB | NPU
63
- | DETR-ResNet50-DC5 | ONNX | float | Qualcomm® QCS8550 (Proxy) | 44.576 ms | 5 - 9 MB | NPU
64
- | DETR-ResNet50-DC5 | ONNX | float | Qualcomm® QCS9075 | 64.459 ms | 5 - 12 MB | NPU
65
- | DETR-ResNet50-DC5 | ONNX | float | Snapdragon® 8 Elite For Galaxy Mobile | 23.566 ms | 3 - 456 MB | NPU
66
- | DETR-ResNet50-DC5 | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 19.02 ms | 5 - 465 MB | NPU
67
- | DETR-ResNet50-DC5 | ONNX | float | Snapdragon® X2 Elite | 20.236 ms | 79 - 79 MB | NPU
68
- | DETR-ResNet50-DC5 | QNN_DLC | float | Snapdragon® X Elite | 47.64 ms | 5 - 5 MB | NPU
69
- | DETR-ResNet50-DC5 | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 33.886 ms | 5 - 631 MB | NPU
70
- | DETR-ResNet50-DC5 | QNN_DLC | float | Qualcomm® QCS8275 (Proxy) | 171.871 ms | 0 - 476 MB | NPU
71
- | DETR-ResNet50-DC5 | QNN_DLC | float | Qualcomm® QCS8550 (Proxy) | 46.552 ms | 5 - 7 MB | NPU
72
- | DETR-ResNet50-DC5 | QNN_DLC | float | Qualcomm® SA8775P | 60.121 ms | 0 - 475 MB | NPU
73
- | DETR-ResNet50-DC5 | QNN_DLC | float | Qualcomm® QCS9075 | 69.625 ms | 5 - 11 MB | NPU
74
- | DETR-ResNet50-DC5 | QNN_DLC | float | Qualcomm® QCS8450 (Proxy) | 74.601 ms | 4 - 448 MB | NPU
75
- | DETR-ResNet50-DC5 | QNN_DLC | float | Qualcomm® SA7255P | 171.871 ms | 0 - 476 MB | NPU
76
- | DETR-ResNet50-DC5 | QNN_DLC | float | Qualcomm® SA8295P | 63.421 ms | 2 - 327 MB | NPU
77
- | DETR-ResNet50-DC5 | QNN_DLC | float | Snapdragon® 8 Elite For Galaxy Mobile | 24.191 ms | 0 - 497 MB | NPU
78
- | DETR-ResNet50-DC5 | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 Mobile | 19.161 ms | 4 - 508 MB | NPU
79
- | DETR-ResNet50-DC5 | QNN_DLC | float | Snapdragon® X2 Elite | 20.812 ms | 5 - 5 MB | NPU
 
 
 
 
 
 
 
 
 
 
80
 
81
  ## License
82
  * The license for the original implementation of DETR-ResNet50-DC5 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-DC5 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_dc5) 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_dc5/releases/v0.48.0/detr_resnet50_dc5-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_dc5/releases/v0.48.0/detr_resnet50_dc5-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_dc5/releases/v0.48.0/detr_resnet50_dc5-tflite-float.zip)
34
 
35
  For more device-specific assets and performance metrics, visit **[DETR-ResNet50-DC5 on Qualcomm® AI Hub](https://aihub.qualcomm.com/models/detr_resnet50_dc5)**.
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_dc5) 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-DC5 on GitHub](https://github.com/qualcomm/ai-hub-models/blob/main/qai_hub_models/models/detr_resnet50_dc5) for usage instructions.
48
 
49
  ## Model Details
50
 
 
58
  ## Performance Summary
59
  | Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit
60
  |---|---|---|---|---|---|---
61
+ | DETR-ResNet50-DC5 | ONNX | float | Snapdragon® X2 Elite | 20.302 ms | 79 - 79 MB | NPU
62
+ | DETR-ResNet50-DC5 | ONNX | float | Snapdragon® X Elite | 45.406 ms | 78 - 78 MB | NPU
63
+ | DETR-ResNet50-DC5 | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 32.511 ms | 2 - 642 MB | NPU
64
+ | DETR-ResNet50-DC5 | ONNX | float | Qualcomm® QCS8550 (Proxy) | 44.622 ms | 0 - 95 MB | NPU
65
+ | DETR-ResNet50-DC5 | ONNX | float | Qualcomm® QCS9075 | 64.675 ms | 5 - 12 MB | NPU
66
+ | DETR-ResNet50-DC5 | ONNX | float | Snapdragon® 8 Elite For Galaxy Mobile | 23.603 ms | 3 - 420 MB | NPU
67
+ | DETR-ResNet50-DC5 | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 18.957 ms | 5 - 474 MB | NPU
68
+ | DETR-ResNet50-DC5 | QNN_DLC | float | Snapdragon® X2 Elite | 20.76 ms | 5 - 5 MB | NPU
69
+ | DETR-ResNet50-DC5 | QNN_DLC | float | Snapdragon® X Elite | 47.595 ms | 5 - 5 MB | NPU
70
+ | DETR-ResNet50-DC5 | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 33.925 ms | 5 - 629 MB | NPU
71
+ | DETR-ResNet50-DC5 | QNN_DLC | float | Qualcomm® QCS8275 (Proxy) | 171.904 ms | 1 - 478 MB | NPU
72
+ | DETR-ResNet50-DC5 | QNN_DLC | float | Qualcomm® QCS8550 (Proxy) | 46.548 ms | 5 - 7 MB | NPU
73
+ | DETR-ResNet50-DC5 | QNN_DLC | float | Qualcomm® SA8775P | 60.101 ms | 1 - 477 MB | NPU
74
+ | DETR-ResNet50-DC5 | QNN_DLC | float | Qualcomm® QCS9075 | 69.754 ms | 5 - 11 MB | NPU
75
+ | DETR-ResNet50-DC5 | QNN_DLC | float | Qualcomm® QCS8450 (Proxy) | 74.725 ms | 3 - 450 MB | NPU
76
+ | DETR-ResNet50-DC5 | QNN_DLC | float | Qualcomm® SA7255P | 171.904 ms | 1 - 478 MB | NPU
77
+ | DETR-ResNet50-DC5 | QNN_DLC | float | Qualcomm® SA8295P | 63.396 ms | 0 - 325 MB | NPU
78
+ | DETR-ResNet50-DC5 | QNN_DLC | float | Snapdragon® 8 Elite For Galaxy Mobile | 24.281 ms | 5 - 502 MB | NPU
79
+ | DETR-ResNet50-DC5 | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 Mobile | 19.169 ms | 5 - 507 MB | NPU
80
+ | DETR-ResNet50-DC5 | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 32.355 ms | 0 - 652 MB | NPU
81
+ | DETR-ResNet50-DC5 | TFLITE | float | Qualcomm® QCS8275 (Proxy) | 168.929 ms | 0 - 515 MB | NPU
82
+ | DETR-ResNet50-DC5 | TFLITE | float | Qualcomm® QCS8550 (Proxy) | 44.04 ms | 0 - 3 MB | NPU
83
+ | DETR-ResNet50-DC5 | TFLITE | float | Qualcomm® SA8775P | 58.046 ms | 0 - 512 MB | NPU
84
+ | DETR-ResNet50-DC5 | TFLITE | float | Qualcomm® QCS9075 | 65.915 ms | 0 - 90 MB | NPU
85
+ | DETR-ResNet50-DC5 | TFLITE | float | Qualcomm® QCS8450 (Proxy) | 79.383 ms | 0 - 477 MB | NPU
86
+ | DETR-ResNet50-DC5 | TFLITE | float | Qualcomm® SA7255P | 168.929 ms | 0 - 515 MB | NPU
87
+ | DETR-ResNet50-DC5 | TFLITE | float | Qualcomm® SA8295P | 65.362 ms | 0 - 382 MB | NPU
88
+ | DETR-ResNet50-DC5 | TFLITE | float | Snapdragon® 8 Elite For Galaxy Mobile | 22.489 ms | 0 - 544 MB | NPU
89
+ | DETR-ResNet50-DC5 | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 17.741 ms | 0 - 536 MB | NPU
90
 
91
  ## License
92
  * The license for the original implementation of DETR-ResNet50-DC5 can be found