Hugging Face's logo

qualcomm
/
SESR-M5

Image-to-Image
PyTorch
LiteRT
ONNX
android
Model card Files Community
SESR-M5
File size: 18,790 Bytes 
af3d024
 
 
 
 
abe750a
af3d024
 
 
04f033f
af3d024
 
 
 
975a374
af3d024
 
6d796e4
975a374
 
af3d024
 
 
 
 
 
 
2c406df
af3d024
88c4457
5c3fbb2
ad1fc07
 
 
88c4457
2c406df
e32c96b
0ceef01
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
ad1fc07
0ceef01
 
ad1fc07
0ceef01
 
ad1fc07
0ceef01
ad1fc07
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
88c4457
af3d024
 
 
 
 
 
1928792
af3d024
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
61cf7d5
e32c96b
 
 
2c406df
 
0ceef01
 
2c406df
 
61cf7d5
88c4457
 
af3d024
 
88c4457
af3d024
 
 
 
 
 
 
 
 
 
 
 
6d78176
af3d024
 
6d78176
af3d024
 
1928792
af3d024
6d78176
 
 
 
 
 
 
 
 
 
 
 
 
 
 
af3d024
 
 
 
 
 
 
 
 
 
 
 
d4a0cbd
 
 
e68e87a
af3d024
 
 
 
 
 
 
 
 
d4a0cbd
 
 
 
e68e87a
af3d024
 
 
 
 
 
04f033f
af3d024
 
 
e68e87a
 
 
 
 
0060e2c
e68e87a
 
 
 
 
0060e2c
e68e87a
 
88c4457
af3d024
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
e32c96b
af3d024
1928792
 
e32c96b
 
 
af3d024
 
 
 
 
e32c96b
 
af3d024
9cd16c2
af3d024
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
 
---
library_name: pytorch
license: other
tags:
- android
pipeline_tag: image-to-image

---

![](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/sesr_m5/web-assets/model_demo.png)

# SESR-M5: Optimized for Mobile Deployment
## Upscale images in real time


SESR M5 performs efficient on-device upscaling of images.

This model is an implementation of SESR-M5 found [here](https://github.com/quic/aimet-model-zoo/tree/develop/aimet_zoo_torch/sesr).


This repository provides scripts to run SESR-M5 on Qualcomm® devices.
More details on model performance across various devices, can be found
[here](https://aihub.qualcomm.com/models/sesr_m5).


### Model Details

- **Model Type:** Model_use_case.super_resolution
- **Model Stats:**
  - Model checkpoint: sesr_m5_3x_checkpoint
  - Input resolution: 128x128
  - Number of parameters: 343K
  - Model size (float): 1.32 MB
  - Model size (w8a8): 395 KB

| Model | Precision | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit | Target Model
|---|---|---|---|---|---|---|---|---|
| SESR-M5 | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 12.601 ms | 6 - 18 MB | NPU | [SESR-M5.tflite](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5.tflite) |
| SESR-M5 | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 10.474 ms | 0 - 12 MB | NPU | [SESR-M5.dlc](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5.dlc) |
| SESR-M5 | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 3.267 ms | 0 - 21 MB | NPU | [SESR-M5.tflite](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5.tflite) |
| SESR-M5 | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 2.799 ms | 0 - 27 MB | NPU | [SESR-M5.dlc](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5.dlc) |
| SESR-M5 | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 2.22 ms | 0 - 7 MB | NPU | [SESR-M5.tflite](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5.tflite) |
| SESR-M5 | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 1.948 ms | 0 - 5 MB | NPU | [SESR-M5.dlc](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5.dlc) |
| SESR-M5 | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 3.767 ms | 0 - 14 MB | NPU | [SESR-M5.tflite](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5.tflite) |
| SESR-M5 | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 3.126 ms | 0 - 14 MB | NPU | [SESR-M5.dlc](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5.dlc) |
| SESR-M5 | float | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 12.601 ms | 6 - 18 MB | NPU | [SESR-M5.tflite](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5.tflite) |
| SESR-M5 | float | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 10.474 ms | 0 - 12 MB | NPU | [SESR-M5.dlc](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5.dlc) |
| SESR-M5 | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 2.181 ms | 0 - 6 MB | NPU | [SESR-M5.tflite](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5.tflite) |
| SESR-M5 | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 1.974 ms | 0 - 6 MB | NPU | [SESR-M5.dlc](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5.dlc) |
| SESR-M5 | float | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 5.541 ms | 0 - 22 MB | NPU | [SESR-M5.tflite](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5.tflite) |
| SESR-M5 | float | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 3.32 ms | 0 - 18 MB | NPU | [SESR-M5.dlc](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5.dlc) |
| SESR-M5 | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 2.216 ms | 0 - 7 MB | NPU | [SESR-M5.tflite](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5.tflite) |
| SESR-M5 | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_DLC | 1.977 ms | 0 - 6 MB | NPU | [SESR-M5.dlc](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5.dlc) |
| SESR-M5 | float | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 3.767 ms | 0 - 14 MB | NPU | [SESR-M5.tflite](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5.tflite) |
| SESR-M5 | float | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 3.126 ms | 0 - 14 MB | NPU | [SESR-M5.dlc](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5.dlc) |
| SESR-M5 | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | TFLITE | 2.325 ms | 0 - 7 MB | NPU | [SESR-M5.tflite](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5.tflite) |
| SESR-M5 | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | QNN_DLC | 1.967 ms | 0 - 6 MB | NPU | [SESR-M5.dlc](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5.dlc) |
| SESR-M5 | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | ONNX | 2.601 ms | 0 - 6 MB | NPU | [SESR-M5.onnx](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5.onnx) |
| SESR-M5 | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 2.388 ms | 0 - 26 MB | NPU | [SESR-M5.tflite](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5.tflite) |
| SESR-M5 | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 1.261 ms | 0 - 22 MB | NPU | [SESR-M5.dlc](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5.dlc) |
| SESR-M5 | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 1.612 ms | 0 - 24 MB | NPU | [SESR-M5.onnx](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5.onnx) |
| SESR-M5 | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | TFLITE | 1.765 ms | 0 - 18 MB | NPU | [SESR-M5.tflite](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5.tflite) |
| SESR-M5 | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | QNN_DLC | 1.304 ms | 0 - 20 MB | NPU | [SESR-M5.dlc](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5.dlc) |
| SESR-M5 | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 1.694 ms | 1 - 15 MB | NPU | [SESR-M5.onnx](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5.onnx) |
| SESR-M5 | float | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 2.192 ms | 0 - 0 MB | NPU | [SESR-M5.dlc](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5.dlc) |
| SESR-M5 | float | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 2.563 ms | 8 - 8 MB | NPU | [SESR-M5.onnx](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5.onnx) |
| SESR-M5 | w8a8 | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 5.802 ms | 2 - 14 MB | NPU | [SESR-M5.tflite](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5_w8a8.tflite) |
| SESR-M5 | w8a8 | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 1.91 ms | 0 - 13 MB | NPU | [SESR-M5.dlc](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5_w8a8.dlc) |
| SESR-M5 | w8a8 | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 1.681 ms | 0 - 22 MB | NPU | [SESR-M5.tflite](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5_w8a8.tflite) |
| SESR-M5 | w8a8 | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 0.938 ms | 0 - 26 MB | NPU | [SESR-M5.dlc](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5_w8a8.dlc) |
| SESR-M5 | w8a8 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 2.302 ms | 1 - 8 MB | NPU | [SESR-M5.tflite](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5_w8a8.tflite) |
| SESR-M5 | w8a8 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 0.633 ms | 0 - 9 MB | NPU | [SESR-M5.dlc](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5_w8a8.dlc) |
| SESR-M5 | w8a8 | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 2.065 ms | 0 - 14 MB | NPU | [SESR-M5.tflite](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5_w8a8.tflite) |
| SESR-M5 | w8a8 | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 0.856 ms | 0 - 15 MB | NPU | [SESR-M5.dlc](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5_w8a8.dlc) |
| SESR-M5 | w8a8 | RB3 Gen 2 (Proxy) | Qualcomm® QCS6490 (Proxy) | TFLITE | 4.085 ms | 2 - 20 MB | NPU | [SESR-M5.tflite](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5_w8a8.tflite) |
| SESR-M5 | w8a8 | RB3 Gen 2 (Proxy) | Qualcomm® QCS6490 (Proxy) | QNN_DLC | 3.015 ms | 0 - 18 MB | NPU | [SESR-M5.dlc](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5_w8a8.dlc) |
| SESR-M5 | w8a8 | RB5 (Proxy) | Qualcomm® QCS8250 (Proxy) | TFLITE | 23.305 ms | 2 - 4 MB | NPU | [SESR-M5.tflite](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5_w8a8.tflite) |
| SESR-M5 | w8a8 | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 5.802 ms | 2 - 14 MB | NPU | [SESR-M5.tflite](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5_w8a8.tflite) |
| SESR-M5 | w8a8 | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 1.91 ms | 0 - 13 MB | NPU | [SESR-M5.dlc](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5_w8a8.dlc) |
| SESR-M5 | w8a8 | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 1.351 ms | 0 - 6 MB | NPU | [SESR-M5.tflite](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5_w8a8.tflite) |
| SESR-M5 | w8a8 | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 0.646 ms | 0 - 10 MB | NPU | [SESR-M5.dlc](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5_w8a8.dlc) |
| SESR-M5 | w8a8 | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 2.361 ms | 0 - 22 MB | NPU | [SESR-M5.tflite](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5_w8a8.tflite) |
| SESR-M5 | w8a8 | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 1.373 ms | 0 - 18 MB | NPU | [SESR-M5.dlc](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5_w8a8.dlc) |
| SESR-M5 | w8a8 | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 1.369 ms | 0 - 6 MB | NPU | [SESR-M5.tflite](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5_w8a8.tflite) |
| SESR-M5 | w8a8 | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_DLC | 0.651 ms | 0 - 9 MB | NPU | [SESR-M5.dlc](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5_w8a8.dlc) |
| SESR-M5 | w8a8 | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 2.065 ms | 0 - 14 MB | NPU | [SESR-M5.tflite](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5_w8a8.tflite) |
| SESR-M5 | w8a8 | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 0.856 ms | 0 - 15 MB | NPU | [SESR-M5.dlc](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5_w8a8.dlc) |
| SESR-M5 | w8a8 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | TFLITE | 1.366 ms | 0 - 7 MB | NPU | [SESR-M5.tflite](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5_w8a8.tflite) |
| SESR-M5 | w8a8 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | QNN_DLC | 0.644 ms | 0 - 10 MB | NPU | [SESR-M5.dlc](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5_w8a8.dlc) |
| SESR-M5 | w8a8 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | ONNX | 0.966 ms | 0 - 8 MB | NPU | [SESR-M5.onnx](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5_w8a8.onnx) |
| SESR-M5 | w8a8 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 1.266 ms | 0 - 25 MB | NPU | [SESR-M5.tflite](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5_w8a8.tflite) |
| SESR-M5 | w8a8 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 0.443 ms | 0 - 26 MB | NPU | [SESR-M5.dlc](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5_w8a8.dlc) |
| SESR-M5 | w8a8 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 0.619 ms | 0 - 28 MB | NPU | [SESR-M5.onnx](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5_w8a8.onnx) |
| SESR-M5 | w8a8 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | TFLITE | 1.602 ms | 0 - 19 MB | NPU | [SESR-M5.tflite](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5_w8a8.tflite) |
| SESR-M5 | w8a8 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | QNN_DLC | 0.461 ms | 0 - 20 MB | NPU | [SESR-M5.dlc](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5_w8a8.dlc) |
| SESR-M5 | w8a8 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 0.695 ms | 1 - 19 MB | NPU | [SESR-M5.onnx](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5_w8a8.onnx) |
| SESR-M5 | w8a8 | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 0.903 ms | 2 - 2 MB | NPU | [SESR-M5.dlc](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5_w8a8.dlc) |
| SESR-M5 | w8a8 | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 0.95 ms | 3 - 3 MB | NPU | [SESR-M5.onnx](https://huggingface.co/qualcomm/SESR-M5/blob/main/SESR-M5_w8a8.onnx) |




## Installation


Install the package via pip:
```bash
pip install qai-hub-models
```


## Configure Qualcomm® AI Hub to run this model on a cloud-hosted device

Sign-in to [Qualcomm® AI Hub](https://app.aihub.qualcomm.com/) with your
Qualcomm® ID. Once signed in navigate to `Account -> Settings -> API Token`.

With this API token, you can configure your client to run models on the cloud
hosted devices.
```bash
qai-hub configure --api_token API_TOKEN
```
Navigate to [docs](https://app.aihub.qualcomm.com/docs/) for more information.



## Demo off target

The package contains a simple end-to-end demo that downloads pre-trained
weights and runs this model on a sample input.

```bash
python -m qai_hub_models.models.sesr_m5.demo
```

The above demo runs a reference implementation of pre-processing, model
inference, and post processing.

**NOTE**: If you want running in a Jupyter Notebook or Google Colab like
environment, please add the following to your cell (instead of the above).
```
%run -m qai_hub_models.models.sesr_m5.demo
```


### Run model on a cloud-hosted device

In addition to the demo, you can also run the model on a cloud-hosted Qualcomm®
device. This script does the following:
* Performance check on-device on a cloud-hosted device
* Downloads compiled assets that can be deployed on-device for Android.
* Accuracy check between PyTorch and on-device outputs.

```bash
python -m qai_hub_models.models.sesr_m5.export
```
```
Profiling Results
------------------------------------------------------------
SESR-M5
Device                          : cs_8275 (ANDROID 14)                
Runtime                         : TFLITE                              
Estimated inference time (ms)   : 12.6                                
Estimated peak memory usage (MB): [6, 18]                             
Total # Ops                     : 25                                  
Compute Unit(s)                 : npu (22 ops) gpu (0 ops) cpu (3 ops)
```


## How does this work?

This [export script](https://aihub.qualcomm.com/models/sesr_m5/qai_hub_models/models/SESR-M5/export.py)
leverages [Qualcomm® AI Hub](https://aihub.qualcomm.com/) to optimize, validate, and deploy this model
on-device. Lets go through each step below in detail:

Step 1: **Compile model for on-device deployment**

To compile a PyTorch model for on-device deployment, we first trace the model
in memory using the `jit.trace` and then call the `submit_compile_job` API.

```python
import torch

import qai_hub as hub
from qai_hub_models.models.sesr_m5 import Model

# Load the model
torch_model = Model.from_pretrained()

# Device
device = hub.Device("Samsung Galaxy S24")

# Trace model
input_shape = torch_model.get_input_spec()
sample_inputs = torch_model.sample_inputs()

pt_model = torch.jit.trace(torch_model, [torch.tensor(data[0]) for _, data in sample_inputs.items()])

# Compile model on a specific device
compile_job = hub.submit_compile_job(
    model=pt_model,
    device=device,
    input_specs=torch_model.get_input_spec(),
)

# Get target model to run on-device
target_model = compile_job.get_target_model()

```


Step 2: **Performance profiling on cloud-hosted device**

After compiling models from step 1. Models can be profiled model on-device using the
`target_model`. Note that this scripts runs the model on a device automatically
provisioned in the cloud.  Once the job is submitted, you can navigate to a
provided job URL to view a variety of on-device performance metrics.
```python
profile_job = hub.submit_profile_job(
    model=target_model,
    device=device,
)
        
```

Step 3: **Verify on-device accuracy**

To verify the accuracy of the model on-device, you can run on-device inference
on sample input data on the same cloud hosted device.
```python
input_data = torch_model.sample_inputs()
inference_job = hub.submit_inference_job(
    model=target_model,
    device=device,
    inputs=input_data,
)
    on_device_output = inference_job.download_output_data()

```
With the output of the model, you can compute like PSNR, relative errors or
spot check the output with expected output.

**Note**: This on-device profiling and inference requires access to Qualcomm®
AI Hub. [Sign up for access](https://myaccount.qualcomm.com/signup).



## Run demo on a cloud-hosted device

You can also run the demo on-device.

```bash
python -m qai_hub_models.models.sesr_m5.demo --eval-mode on-device
```

**NOTE**: If you want running in a Jupyter Notebook or Google Colab like
environment, please add the following to your cell (instead of the above).
```
%run -m qai_hub_models.models.sesr_m5.demo -- --eval-mode on-device
```


## Deploying compiled model to Android


The models can be deployed using multiple runtimes:
- TensorFlow Lite (`.tflite` export): [This
  tutorial](https://www.tensorflow.org/lite/android/quickstart) provides a
  guide to deploy the .tflite model in an Android application.


- QNN (`.so` export ): This [sample
  app](https://docs.qualcomm.com/bundle/publicresource/topics/80-63442-50/sample_app.html)
provides instructions on how to use the `.so` shared library  in an Android application.


## View on Qualcomm® AI Hub
Get more details on SESR-M5's performance across various devices [here](https://aihub.qualcomm.com/models/sesr_m5).
Explore all available models on [Qualcomm® AI Hub](https://aihub.qualcomm.com/)


## License
* The license for the original implementation of SESR-M5 can be found
  [here](https://github.com/quic/aimet-model-zoo/blob/develop/LICENSE.pdf).
* 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)



## References
* [Collapsible Linear Blocks for Super-Efficient Super Resolution](https://arxiv.org/abs/2103.09404)
* [Source Model Implementation](https://github.com/quic/aimet-model-zoo/tree/develop/aimet_zoo_torch/sesr)



## Community
* Join [our AI Hub Slack community](https://aihub.qualcomm.com/community/slack) to collaborate, post questions and learn more about on-device AI.
* For questions or feedback please [reach out to us](mailto:[email protected]).