Hugging Face's logo

qualcomm
/
Real-ESRGAN-x4plus

Image-to-Image
PyTorch
TF Lite
ONNX
android
Model card Files Community
1
qaihm-bot commited on
Commit
bc6430a
•
1 Parent(s): 40ce7ed

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +40 -19
README.md CHANGED
@@ -14,7 +14,7 @@ tags:
14
 
15
  Real-ESRGAN is a machine learning model that upscales an image with minimal loss in quality. The implementation is a derivative of the Real-ESRGAN-x4plus architecture, a larger and more powerful version compared to the Real-ESRGAN-general-x4v3 architecture.
16
 
17
- This model is an implementation of Real-ESRGAN-x4plus found [here](https://github.com/xinntao/Real-ESRGAN).
18
  This repository provides scripts to run Real-ESRGAN-x4plus on Qualcomm® devices.
19
  More details on model performance across various devices, can be found
20
  [here](https://aihub.qualcomm.com/models/real_esrgan_x4plus).
@@ -29,15 +29,32 @@ More details on model performance across various devices, can be found
29
  - Model checkpoint: RealESRGAN_x4plus
30
  - Input resolution: 128x128
31
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
32
 
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 | 67.256 ms | 3 - 11 MB | FP16 | NPU | [Real-ESRGAN-x4plus.tflite](https://huggingface.co/qualcomm/Real-ESRGAN-x4plus/blob/main/Real-ESRGAN-x4plus.tflite)
38
- | Samsung Galaxy S23 Ultra (Android 13) | Snapdragon® 8 Gen 2 | QNN Model Library | 70.685 ms | 0 - 35 MB | FP16 | NPU | [Real-ESRGAN-x4plus.so](https://huggingface.co/qualcomm/Real-ESRGAN-x4plus/blob/main/Real-ESRGAN-x4plus.so)
39
-
40
-
41
 
42
  ## Installation
43
 
@@ -93,16 +110,16 @@ device. This script does the following:
93
  ```bash
94
  python -m qai_hub_models.models.real_esrgan_x4plus.export
95
  ```
96
-
97
  ```
98
- Profile Job summary of Real-ESRGAN-x4plus
99
- --------------------------------------------------
100
- Device: Snapdragon X Elite CRD (11)
101
- Estimated Inference Time: 65.09 ms
102
- Estimated Peak Memory Range: 0.20-0.20 MB
103
- Compute Units: NPU (1029) | Total (1029)
104
-
105
-
 
106
  ```
107
 
108
 
@@ -201,15 +218,19 @@ provides instructions on how to use the `.so` shared library in an Android appl
201
  Get more details on Real-ESRGAN-x4plus's performance across various devices [here](https://aihub.qualcomm.com/models/real_esrgan_x4plus).
202
  Explore all available models on [Qualcomm® AI Hub](https://aihub.qualcomm.com/)
203
 
 
204
  ## License
205
- - The license for the original implementation of Real-ESRGAN-x4plus can be found
206
- [here](https://github.com/xinntao/Real-ESRGAN/blob/master/LICENSE).
207
- - 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)
 
208
 
209
  ## References
210
  * [Real-ESRGAN: Training Real-World Blind Super-Resolution with Pure Synthetic Data](https://arxiv.org/abs/2107.10833)
211
  * [Source Model Implementation](https://github.com/xinntao/Real-ESRGAN)
212
 
 
 
213
  ## Community
214
  * Join [our AI Hub Slack community](https://aihub.qualcomm.com/community/slack) to collaborate, post questions and learn more about on-device AI.
215
  * For questions or feedback please [reach out to us](mailto:[email protected]).
 
14
 
15
  Real-ESRGAN is a machine learning model that upscales an image with minimal loss in quality. The implementation is a derivative of the Real-ESRGAN-x4plus architecture, a larger and more powerful version compared to the Real-ESRGAN-general-x4v3 architecture.
16
 
17
+ This model is an implementation of Real-ESRGAN-x4plus found [here]({source_repo}).
18
  This repository provides scripts to run Real-ESRGAN-x4plus on Qualcomm® devices.
19
  More details on model performance across various devices, can be found
20
  [here](https://aihub.qualcomm.com/models/real_esrgan_x4plus).
 
29
  - Model checkpoint: RealESRGAN_x4plus
30
  - Input resolution: 128x128
31
 
32
+ | Model | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Precision | Primary Compute Unit | Target Model
33
+ |---|---|---|---|---|---|---|---|---|
34
+ | Real-ESRGAN-x4plus | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | TFLITE | 68.798 ms | 3 - 6 MB | FP16 | NPU | [Real-ESRGAN-x4plus.tflite](https://huggingface.co/qualcomm/Real-ESRGAN-x4plus/blob/main/Real-ESRGAN-x4plus.tflite) |
35
+ | Real-ESRGAN-x4plus | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | QNN | 67.503 ms | 0 - 109 MB | FP16 | NPU | [Real-ESRGAN-x4plus.so](https://huggingface.co/qualcomm/Real-ESRGAN-x4plus/blob/main/Real-ESRGAN-x4plus.so) |
36
+ | Real-ESRGAN-x4plus | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | ONNX | 70.73 ms | 0 - 43 MB | FP16 | NPU | [Real-ESRGAN-x4plus.onnx](https://huggingface.co/qualcomm/Real-ESRGAN-x4plus/blob/main/Real-ESRGAN-x4plus.onnx) |
37
+ | Real-ESRGAN-x4plus | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | TFLITE | 55.834 ms | 3 - 662 MB | FP16 | NPU | [Real-ESRGAN-x4plus.tflite](https://huggingface.co/qualcomm/Real-ESRGAN-x4plus/blob/main/Real-ESRGAN-x4plus.tflite) |
38
+ | Real-ESRGAN-x4plus | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | QNN | 55.888 ms | 0 - 108 MB | FP16 | NPU | [Real-ESRGAN-x4plus.so](https://huggingface.co/qualcomm/Real-ESRGAN-x4plus/blob/main/Real-ESRGAN-x4plus.so) |
39
+ | Real-ESRGAN-x4plus | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | ONNX | 55.527 ms | 8 - 697 MB | FP16 | NPU | [Real-ESRGAN-x4plus.onnx](https://huggingface.co/qualcomm/Real-ESRGAN-x4plus/blob/main/Real-ESRGAN-x4plus.onnx) |
40
+ | Real-ESRGAN-x4plus | QCS8550 (Proxy) | QCS8550 Proxy | TFLITE | 61.376 ms | 1 - 4 MB | FP16 | NPU | [Real-ESRGAN-x4plus.tflite](https://huggingface.co/qualcomm/Real-ESRGAN-x4plus/blob/main/Real-ESRGAN-x4plus.tflite) |
41
+ | Real-ESRGAN-x4plus | QCS8550 (Proxy) | QCS8550 Proxy | QNN | 62.924 ms | 0 - 2 MB | FP16 | NPU | Use Export Script |
42
+ | Real-ESRGAN-x4plus | SA8255 (Proxy) | SA8255P Proxy | TFLITE | 66.879 ms | 3 - 5 MB | FP16 | NPU | [Real-ESRGAN-x4plus.tflite](https://huggingface.co/qualcomm/Real-ESRGAN-x4plus/blob/main/Real-ESRGAN-x4plus.tflite) |
43
+ | Real-ESRGAN-x4plus | SA8255 (Proxy) | SA8255P Proxy | QNN | 63.674 ms | 0 - 2 MB | FP16 | NPU | Use Export Script |
44
+ | Real-ESRGAN-x4plus | SA8775 (Proxy) | SA8775P Proxy | TFLITE | 63.276 ms | 3 - 6 MB | FP16 | NPU | [Real-ESRGAN-x4plus.tflite](https://huggingface.co/qualcomm/Real-ESRGAN-x4plus/blob/main/Real-ESRGAN-x4plus.tflite) |
45
+ | Real-ESRGAN-x4plus | SA8775 (Proxy) | SA8775P Proxy | QNN | 63.755 ms | 0 - 2 MB | FP16 | NPU | Use Export Script |
46
+ | Real-ESRGAN-x4plus | SA8650 (Proxy) | SA8650P Proxy | TFLITE | 66.934 ms | 3 - 6 MB | FP16 | NPU | [Real-ESRGAN-x4plus.tflite](https://huggingface.co/qualcomm/Real-ESRGAN-x4plus/blob/main/Real-ESRGAN-x4plus.tflite) |
47
+ | Real-ESRGAN-x4plus | SA8650 (Proxy) | SA8650P Proxy | QNN | 63.721 ms | 0 - 2 MB | FP16 | NPU | Use Export Script |
48
+ | Real-ESRGAN-x4plus | QCS8450 (Proxy) | QCS8450 Proxy | TFLITE | 143.595 ms | 0 - 617 MB | FP16 | NPU | [Real-ESRGAN-x4plus.tflite](https://huggingface.co/qualcomm/Real-ESRGAN-x4plus/blob/main/Real-ESRGAN-x4plus.tflite) |
49
+ | Real-ESRGAN-x4plus | QCS8450 (Proxy) | QCS8450 Proxy | QNN | 138.583 ms | 0 - 85 MB | FP16 | NPU | Use Export Script |
50
+ | Real-ESRGAN-x4plus | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | TFLITE | 42.951 ms | 3 - 184 MB | FP16 | NPU | [Real-ESRGAN-x4plus.tflite](https://huggingface.co/qualcomm/Real-ESRGAN-x4plus/blob/main/Real-ESRGAN-x4plus.tflite) |
51
+ | Real-ESRGAN-x4plus | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | QNN | 43.454 ms | 0 - 129 MB | FP16 | NPU | Use Export Script |
52
+ | Real-ESRGAN-x4plus | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | ONNX | 43.103 ms | 0 - 177 MB | FP16 | NPU | [Real-ESRGAN-x4plus.onnx](https://huggingface.co/qualcomm/Real-ESRGAN-x4plus/blob/main/Real-ESRGAN-x4plus.onnx) |
53
+ | Real-ESRGAN-x4plus | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN | 65.203 ms | 0 - 0 MB | FP16 | NPU | Use Export Script |
54
+ | Real-ESRGAN-x4plus | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 65.666 ms | 39 - 39 MB | FP16 | NPU | [Real-ESRGAN-x4plus.onnx](https://huggingface.co/qualcomm/Real-ESRGAN-x4plus/blob/main/Real-ESRGAN-x4plus.onnx) |
55
 
56
 
57
 
 
 
 
 
 
 
58
 
59
  ## Installation
60
 
 
110
  ```bash
111
  python -m qai_hub_models.models.real_esrgan_x4plus.export
112
  ```
 
113
  ```
114
+ Profiling Results
115
+ ------------------------------------------------------------
116
+ Real-ESRGAN-x4plus
117
+ Device : Samsung Galaxy S23 (13)
118
+ Runtime : TFLITE
119
+ Estimated inference time (ms) : 68.8
120
+ Estimated peak memory usage (MB): [3, 6]
121
+ Total # Ops : 1028
122
+ Compute Unit(s) : NPU (1028 ops)
123
  ```
124
 
125
 
 
218
  Get more details on Real-ESRGAN-x4plus's performance across various devices [here](https://aihub.qualcomm.com/models/real_esrgan_x4plus).
219
  Explore all available models on [Qualcomm® AI Hub](https://aihub.qualcomm.com/)
220
 
221
+
222
  ## License
223
+ * The license for the original implementation of Real-ESRGAN-x4plus can be found [here](https://github.com/xinntao/Real-ESRGAN/blob/master/LICENSE).
224
+ * 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)
225
+
226
+
227
 
228
  ## References
229
  * [Real-ESRGAN: Training Real-World Blind Super-Resolution with Pure Synthetic Data](https://arxiv.org/abs/2107.10833)
230
  * [Source Model Implementation](https://github.com/xinntao/Real-ESRGAN)
231
 
232
+
233
+
234
  ## Community
235
  * Join [our AI Hub Slack community](https://aihub.qualcomm.com/community/slack) to collaborate, post questions and learn more about on-device AI.
236
  * For questions or feedback please [reach out to us](mailto:[email protected]).