Hugging Face's logo

qualcomm
/
AOT-GAN

Image-to-Image
PyTorch
android
Model card Files
xet
 Community
1
qaihm-bot commited on
Commit
03f26a1
·
verified ·
1 Parent(s): ea3d8b0

v0.60.0

Browse files

See https://github.com/qualcomm/ai-hub-models/releases/v0.60.0 for changelog.

Files changed (2) hide show
  1. README.md +51 -50
  2. release_assets.json +4 -4
README.md CHANGED
@@ -14,7 +14,7 @@ pipeline_tag: image-to-image
14
  AOT-GAN is a machine learning model that allows to erase and in-paint part of given input image.
15
 
16
  This is based on the implementation of AOT-GAN found [here](https://github.com/researchmm/AOT-GAN-for-Inpainting).
17
- 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/v0.59.0/src/qai_hub_models/models/aotgan) library to export with custom configurations. More details on model performance across various devices, can be found [here](#performance-summary).
18
 
19
  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.
20
 
@@ -27,79 +27,80 @@ Below are pre-exported model assets ready for deployment.
27
 
28
  | Runtime | Precision | Chipset | SDK Versions | Download |
29
  |---|---|---|---|---|
30
- | ONNX | float | Universal | QAIRT 2.45, ONNX Runtime 1.27.1 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/aotgan/releases/v0.59.0/aotgan-onnx-float.zip)
31
- | QNN_DLC | float | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/aotgan/releases/v0.59.0/aotgan-qnn_dlc-float.zip)
32
- | TFLITE | float | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/aotgan/releases/v0.59.0/aotgan-tflite-float.zip)
33
 
34
  For more device-specific assets and performance metrics, visit **[AOT-GAN on Qualcomm® AI Hub](https://aihub.qualcomm.com/models/aotgan)**.
35
 
36
 
37
  ### Option 2: Export with Custom Configurations
38
 
39
- Use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/v0.59.0/src/qai_hub_models/models/aotgan) Python library to compile and export the model with your own:
40
  - Custom weights (e.g., fine-tuned checkpoints)
41
  - Custom input shapes
42
  - Target device and runtime configurations
43
 
44
  This option is ideal if you need to customize the model beyond the default configuration provided here.
45
 
46
- See our repository for [AOT-GAN on GitHub](https://github.com/qualcomm/ai-hub-models/blob/v0.59.0/src/qai_hub_models/models/aotgan) for usage instructions.
47
 
48
  ## Model Details
49
 
50
  **Model Type:** Model_use_case.image_editing
51
 
52
  **Model Stats:**
53
- - Model checkpoint: CelebAHQ
54
  - Input resolution: 512x512
55
- - Number of parameters: 15.2M
56
  - Model size (float): 58.0 MB
 
57
 
58
  ## Performance Summary
59
  | Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit
60
  |---|---|---|---|---|---|---
61
- | AOT-GAN | ONNX | float | Snapdragon® X2 Elite | 54.827 ms | 8 - 8 MB | NPU
62
- | AOT-GAN | ONNX | float | Snapdragon® X Elite | 138.511 ms | 32 - 32 MB | NPU
63
- | AOT-GAN | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 96.111 ms | 0 - 708 MB | NPU
64
- | AOT-GAN | ONNX | float | Snapdragon® 8 Gen 1 Mobile | 208.859 ms | 11 - 614 MB | NPU
65
- | AOT-GAN | ONNX | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 134.616 ms | 0 - 39 MB | NPU
66
- | AOT-GAN | ONNX | float | Qualcomm® QCS8450 | 208.859 ms | 11 - 614 MB | NPU
67
- | AOT-GAN | ONNX | float | Qualcomm® Dragonwing™ IQ-9075 | 167.44 ms | 4 - 11 MB | NPU
68
- | AOT-GAN | ONNX | float | Qualcomm® Dragonwing™ IQ-X7181 | 138.511 ms | 32 - 32 MB | NPU
69
- | AOT-GAN | ONNX | float | Qualcomm® Dragonwing™ Q-8750 | 75.372 ms | 8 - 617 MB | NPU
70
- | AOT-GAN | ONNX | float | Snapdragon® 8 Elite Mobile | 75.372 ms | 8 - 617 MB | NPU
71
- | AOT-GAN | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 51.051 ms | 5 - 508 MB | NPU
72
- | AOT-GAN | QNN_DLC | float | Snapdragon® X2 Elite | 50.961 ms | 4 - 4 MB | NPU
73
- | AOT-GAN | QNN_DLC | float | Snapdragon® X Elite | 122.244 ms | 4 - 4 MB | NPU
74
- | AOT-GAN | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 88.438 ms | 1 - 692 MB | NPU
75
- | AOT-GAN | QNN_DLC | float | Snapdragon® 8 Gen 1 Mobile | 200.081 ms | 3 - 605 MB | NPU
76
- | AOT-GAN | QNN_DLC | float | Qualcomm® Dragonwing™ QCS8275 | 540.929 ms | 1 - 544 MB | NPU
77
- | AOT-GAN | QNN_DLC | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 122.345 ms | 4 - 6 MB | NPU
78
- | AOT-GAN | QNN_DLC | float | Qualcomm® SA8775P | 161.29 ms | 1 - 544 MB | NPU
79
- | AOT-GAN | QNN_DLC | float | Qualcomm® SA8650P | 161.29 ms | 1 - 544 MB | NPU
80
- | AOT-GAN | QNN_DLC | float | Qualcomm® SA8255P | 161.29 ms | 1 - 544 MB | NPU
81
- | AOT-GAN | QNN_DLC | float | Qualcomm® QCS8450 | 200.081 ms | 3 - 605 MB | NPU
82
- | AOT-GAN | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-9075 | 160.813 ms | 5 - 14 MB | NPU
83
- | AOT-GAN | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-X7181 | 122.244 ms | 4 - 4 MB | NPU
84
- | AOT-GAN | QNN_DLC | float | Qualcomm® Dragonwing™ Q-8750 | 69.464 ms | 1 - 572 MB | NPU
85
- | AOT-GAN | QNN_DLC | float | Qualcomm® SA7255P | 540.929 ms | 1 - 544 MB | NPU
86
- | AOT-GAN | QNN_DLC | float | Qualcomm® SA8295P | 178.691 ms | 1 - 476 MB | NPU
87
- | AOT-GAN | QNN_DLC | float | Snapdragon® 8 Elite Mobile | 69.464 ms | 1 - 572 MB | NPU
88
- | AOT-GAN | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 Mobile | 46.924 ms | 3 - 490 MB | NPU
89
- | AOT-GAN | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 88.282 ms | 0 - 715 MB | NPU
90
- | AOT-GAN | TFLITE | float | Snapdragon® 8 Gen 1 Mobile | 198.292 ms | 4 - 639 MB | NPU
91
- | AOT-GAN | TFLITE | float | Qualcomm® Dragonwing™ QCS8275 | 541.182 ms | 3 - 558 MB | NPU
92
- | AOT-GAN | TFLITE | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 119.917 ms | 3 - 13 MB | NPU
93
- | AOT-GAN | TFLITE | float | Qualcomm® SA8775P | 161.473 ms | 3 - 557 MB | NPU
94
- | AOT-GAN | TFLITE | float | Qualcomm® SA8650P | 161.473 ms | 3 - 557 MB | NPU
95
- | AOT-GAN | TFLITE | float | Qualcomm® SA8255P | 161.473 ms | 3 - 557 MB | NPU
96
- | AOT-GAN | TFLITE | float | Qualcomm® QCS8450 | 198.292 ms | 4 - 639 MB | NPU
97
- | AOT-GAN | TFLITE | float | Qualcomm® Dragonwing™ IQ-9075 | 161.187 ms | 3 - 48 MB | NPU
98
- | AOT-GAN | TFLITE | float | Qualcomm® Dragonwing™ Q-8750 | 69.199 ms | 2 - 590 MB | NPU
99
- | AOT-GAN | TFLITE | float | Qualcomm® SA7255P | 541.182 ms | 3 - 558 MB | NPU
100
- | AOT-GAN | TFLITE | float | Qualcomm® SA8295P | 178.613 ms | 3 - 494 MB | NPU
101
- | AOT-GAN | TFLITE | float | Snapdragon® 8 Elite Mobile | 69.199 ms | 2 - 590 MB | NPU
102
- | AOT-GAN | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 47.276 ms | 2 - 512 MB | NPU
 
103
 
104
  ## License
105
  * The license for the original implementation of AOT-GAN can be found
 
14
  AOT-GAN is a machine learning model that allows to erase and in-paint part of given input image.
15
 
16
  This is based on the implementation of AOT-GAN found [here](https://github.com/researchmm/AOT-GAN-for-Inpainting).
17
+ 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/v0.60.0/src/qai_hub_models/models/aotgan) library to export with custom configurations. More details on model performance across various devices, can be found [here](#performance-summary).
18
 
19
  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.
20
 
 
27
 
28
  | Runtime | Precision | Chipset | SDK Versions | Download |
29
  |---|---|---|---|---|
30
+ | ONNX | float | Universal | QAIRT 2.45, ONNX Runtime 1.27.1 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/aotgan/releases/v0.60.0/aotgan-onnx-float.zip)
31
+ | QNN_DLC | float | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/aotgan/releases/v0.60.0/aotgan-qnn_dlc-float.zip)
32
+ | TFLITE | float | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/aotgan/releases/v0.60.0/aotgan-tflite-float.zip)
33
 
34
  For more device-specific assets and performance metrics, visit **[AOT-GAN on Qualcomm® AI Hub](https://aihub.qualcomm.com/models/aotgan)**.
35
 
36
 
37
  ### Option 2: Export with Custom Configurations
38
 
39
+ Use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/v0.60.0/src/qai_hub_models/models/aotgan) Python library to compile and export the model with your own:
40
  - Custom weights (e.g., fine-tuned checkpoints)
41
  - Custom input shapes
42
  - Target device and runtime configurations
43
 
44
  This option is ideal if you need to customize the model beyond the default configuration provided here.
45
 
46
+ See our repository for [AOT-GAN on GitHub](https://github.com/qualcomm/ai-hub-models/blob/v0.60.0/src/qai_hub_models/models/aotgan) for usage instructions.
47
 
48
  ## Model Details
49
 
50
  **Model Type:** Model_use_case.image_editing
51
 
52
  **Model Stats:**
 
53
  - Input resolution: 512x512
54
+ - Model checkpoint: CelebAHQ
55
  - Model size (float): 58.0 MB
56
+ - Number of parameters: 15.2M
57
 
58
  ## Performance Summary
59
  | Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit
60
  |---|---|---|---|---|---|---
61
+ | AOT-GAN | ONNX | float | Snapdragon® X2 Elite | 71.914 ms | 8 - 8 MB | NPU
62
+ | AOT-GAN | ONNX | float | Snapdragon® X Elite | 138.439 ms | 32 - 32 MB | NPU
63
+ | AOT-GAN | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 95.569 ms | 11 - 719 MB | NPU
64
+ | AOT-GAN | ONNX | float | Snapdragon® 8 Gen 1 Mobile | 208.331 ms | 1 - 602 MB | NPU
65
+ | AOT-GAN | ONNX | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 134.823 ms | 0 - 39 MB | NPU
66
+ | AOT-GAN | ONNX | float | Qualcomm® QCS8450 | 208.331 ms | 1 - 602 MB | NPU
67
+ | AOT-GAN | ONNX | float | Qualcomm® Dragonwing™ IQ-9075 | 167.1 ms | 4 - 11 MB | NPU
68
+ | AOT-GAN | ONNX | float | Qualcomm® Dragonwing™ IQ-X7181 | 138.439 ms | 32 - 32 MB | NPU
69
+ | AOT-GAN | ONNX | float | Qualcomm® Dragonwing™ Q-8750 | 75.327 ms | 8 - 627 MB | NPU
70
+ | AOT-GAN | ONNX | float | Snapdragon® 8 Elite Mobile | 75.327 ms | 8 - 627 MB | NPU
71
+ | AOT-GAN | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 51.631 ms | 5 - 510 MB | NPU
72
+ | AOT-GAN | QNN_DLC | float | Snapdragon® X2 Elite | 51.047 ms | 4 - 4 MB | NPU
73
+ | AOT-GAN | QNN_DLC | float | Snapdragon® X Elite | 122.254 ms | 4 - 4 MB | NPU
74
+ | AOT-GAN | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 88.731 ms | 0 - 691 MB | NPU
75
+ | AOT-GAN | QNN_DLC | float | Snapdragon® 8 Gen 1 Mobile | 198.46 ms | 3 - 608 MB | NPU
76
+ | AOT-GAN | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-8275 | 541.083 ms | 1 - 543 MB | NPU
77
+ | AOT-GAN | QNN_DLC | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 120.984 ms | 4 - 6 MB | NPU
78
+ | AOT-GAN | QNN_DLC | float | Qualcomm® SA8775P | 161.266 ms | 1 - 543 MB | NPU
79
+ | AOT-GAN | QNN_DLC | float | Qualcomm® SA8650P | 161.266 ms | 1 - 543 MB | NPU
80
+ | AOT-GAN | QNN_DLC | float | Qualcomm® SA8255P | 161.266 ms | 1 - 543 MB | NPU
81
+ | AOT-GAN | QNN_DLC | float | Qualcomm® QCS8450 | 198.46 ms | 3 - 608 MB | NPU
82
+ | AOT-GAN | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-9075 | 160.879 ms | 4 - 13 MB | NPU
83
+ | AOT-GAN | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-X7181 | 122.254 ms | 4 - 4 MB | NPU
84
+ | AOT-GAN | QNN_DLC | float | Qualcomm® Dragonwing™ Q-8750 | 69.611 ms | 3 - 574 MB | NPU
85
+ | AOT-GAN | QNN_DLC | float | Qualcomm® SA7255P | 541.083 ms | 1 - 543 MB | NPU
86
+ | AOT-GAN | QNN_DLC | float | Qualcomm® SA8295P | 178.723 ms | 1 - 476 MB | NPU
87
+ | AOT-GAN | QNN_DLC | float | Snapdragon® 8 Elite Mobile | 69.611 ms | 3 - 574 MB | NPU
88
+ | AOT-GAN | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 Mobile | 47.001 ms | 2 - 491 MB | NPU
89
+ | AOT-GAN | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 88.383 ms | 2 - 717 MB | NPU
90
+ | AOT-GAN | TFLITE | float | Snapdragon® 8 Gen 1 Mobile | 198.551 ms | 4 - 645 MB | NPU
91
+ | AOT-GAN | TFLITE | float | Qualcomm® Dragonwing™ IQ-8275 | 175.52 ms | 3 - 48 MB | NPU
92
+ | AOT-GAN | TFLITE | float | Qualcomm® Dragonwing™ IQ-8275 | 541.494 ms | 3 - 557 MB | NPU
93
+ | AOT-GAN | TFLITE | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 120.046 ms | 3 - 6 MB | NPU
94
+ | AOT-GAN | TFLITE | float | Qualcomm® SA8775P | 161.48 ms | 3 - 556 MB | NPU
95
+ | AOT-GAN | TFLITE | float | Qualcomm® SA8650P | 161.48 ms | 3 - 556 MB | NPU
96
+ | AOT-GAN | TFLITE | float | Qualcomm® SA8255P | 161.48 ms | 3 - 556 MB | NPU
97
+ | AOT-GAN | TFLITE | float | Qualcomm® QCS8450 | 198.551 ms | 4 - 645 MB | NPU
98
+ | AOT-GAN | TFLITE | float | Qualcomm® Dragonwing™ IQ-9075 | 161.529 ms | 3 - 48 MB | NPU
99
+ | AOT-GAN | TFLITE | float | Qualcomm® Dragonwing™ Q-8750 | 69.134 ms | 3 - 590 MB | NPU
100
+ | AOT-GAN | TFLITE | float | Qualcomm® SA7255P | 541.494 ms | 3 - 557 MB | NPU
101
+ | AOT-GAN | TFLITE | float | Qualcomm® SA8295P | 178.664 ms | 4 - 494 MB | NPU
102
+ | AOT-GAN | TFLITE | float | Snapdragon® 8 Elite Mobile | 69.134 ms | 3 - 590 MB | NPU
103
+ | AOT-GAN | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 47.148 ms | 2 - 511 MB | NPU
104
 
105
  ## License
106
  * The license for the original implementation of AOT-GAN can be found
release_assets.json CHANGED
@@ -1,5 +1,5 @@
1
  {
2
- "version": "0.59.0",
3
  "precisions": {
4
  "float": {
5
  "universal_assets": {
@@ -8,19 +8,19 @@
8
  "qairt": "2.45.0.260326154327",
9
  "onnx_runtime": "1.27.1"
10
  },
11
- "download_url": "https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/aotgan/releases/v0.59.0/aotgan-onnx-float.zip"
12
  },
13
  "qnn_dlc": {
14
  "tool_versions": {
15
  "qairt": "2.45.0.260326154327"
16
  },
17
- "download_url": "https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/aotgan/releases/v0.59.0/aotgan-qnn_dlc-float.zip"
18
  },
19
  "tflite": {
20
  "tool_versions": {
21
  "qairt": "2.45.0.260326154327"
22
  },
23
- "download_url": "https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/aotgan/releases/v0.59.0/aotgan-tflite-float.zip"
24
  }
25
  }
26
  }
 
1
  {
2
+ "version": "0.60.0",
3
  "precisions": {
4
  "float": {
5
  "universal_assets": {
 
8
  "qairt": "2.45.0.260326154327",
9
  "onnx_runtime": "1.27.1"
10
  },
11
+ "download_url": "https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/aotgan/releases/v0.60.0/aotgan-onnx-float.zip"
12
  },
13
  "qnn_dlc": {
14
  "tool_versions": {
15
  "qairt": "2.45.0.260326154327"
16
  },
17
+ "download_url": "https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/aotgan/releases/v0.60.0/aotgan-qnn_dlc-float.zip"
18
  },
19
  "tflite": {
20
  "tool_versions": {
21
  "qairt": "2.45.0.260326154327"
22
  },
23
+ "download_url": "https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/aotgan/releases/v0.60.0/aotgan-tflite-float.zip"
24
  }
25
  }
26
  }