qaihm-bot commited on
Commit
c00f6d5
·
verified ·
1 Parent(s): 15e3db3

See https://github.com/quic/ai-hub-models/releases/v0.30.5 for changelog.

Files changed (3) hide show
  1. README.md +36 -37
  2. Swin-Small.onnx +2 -2
  3. Swin-Small_w8a16.onnx +2 -2
README.md CHANGED
@@ -35,39 +35,38 @@ More details on model performance across various devices, can be found
35
 
36
  | Model | Precision | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit | Target Model
37
  |---|---|---|---|---|---|---|---|---|
38
- | Swin-Small | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 44.455 ms | 0 - 258 MB | NPU | [Swin-Small.tflite](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.tflite) |
39
- | Swin-Small | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN | 41.812 ms | 1 - 10 MB | NPU | Use Export Script |
40
- | Swin-Small | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 23.285 ms | 0 - 261 MB | NPU | [Swin-Small.tflite](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.tflite) |
41
- | Swin-Small | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN | 25.952 ms | 0 - 245 MB | NPU | Use Export Script |
42
- | Swin-Small | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 18.353 ms | 0 - 29 MB | NPU | [Swin-Small.tflite](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.tflite) |
43
- | Swin-Small | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN | 17.569 ms | 1 - 4 MB | NPU | Use Export Script |
44
- | Swin-Small | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 20.715 ms | 0 - 258 MB | NPU | [Swin-Small.tflite](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.tflite) |
45
- | Swin-Small | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN | 19.597 ms | 1 - 12 MB | NPU | Use Export Script |
46
- | Swin-Small | float | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 44.455 ms | 0 - 258 MB | NPU | [Swin-Small.tflite](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.tflite) |
47
- | Swin-Small | float | SA7255P ADP | Qualcomm® SA7255P | QNN | 41.812 ms | 1 - 10 MB | NPU | Use Export Script |
48
- | Swin-Small | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 18.339 ms | 0 - 29 MB | NPU | [Swin-Small.tflite](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.tflite) |
49
- | Swin-Small | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN | 17.666 ms | 1 - 3 MB | NPU | Use Export Script |
50
- | Swin-Small | float | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 25.451 ms | 0 - 251 MB | NPU | [Swin-Small.tflite](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.tflite) |
51
- | Swin-Small | float | SA8295P ADP | Qualcomm® SA8295P | QNN | 24.516 ms | 1 - 18 MB | NPU | Use Export Script |
52
- | Swin-Small | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 18.377 ms | 0 - 29 MB | NPU | [Swin-Small.tflite](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.tflite) |
53
- | Swin-Small | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN | 17.6 ms | 1 - 3 MB | NPU | Use Export Script |
54
- | Swin-Small | float | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 20.715 ms | 0 - 258 MB | NPU | [Swin-Small.tflite](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.tflite) |
55
- | Swin-Small | float | SA8775P ADP | Qualcomm® SA8775P | QNN | 19.597 ms | 1 - 12 MB | NPU | Use Export Script |
56
- | Swin-Small | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | TFLITE | 18.399 ms | 0 - 24 MB | NPU | [Swin-Small.tflite](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.tflite) |
57
- | Swin-Small | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | QNN | 17.589 ms | 0 - 61 MB | NPU | Use Export Script |
58
- | Swin-Small | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | ONNX | 16.511 ms | 0 - 278 MB | NPU | [Swin-Small.onnx](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.onnx) |
59
- | Swin-Small | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 12.542 ms | 0 - 261 MB | NPU | [Swin-Small.tflite](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.tflite) |
60
- | Swin-Small | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN | 11.918 ms | 1 - 251 MB | NPU | Use Export Script |
61
- | Swin-Small | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 11.255 ms | 1 - 248 MB | NPU | [Swin-Small.onnx](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.onnx) |
62
- | Swin-Small | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | TFLITE | 11.82 ms | 0 - 258 MB | NPU | [Swin-Small.tflite](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.tflite) |
63
- | Swin-Small | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | QNN | 10.947 ms | 1 - 242 MB | NPU | Use Export Script |
64
- | Swin-Small | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 10.263 ms | 1 - 233 MB | NPU | [Swin-Small.onnx](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.onnx) |
65
- | Swin-Small | float | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN | 18.223 ms | 1 - 1 MB | NPU | Use Export Script |
66
- | Swin-Small | float | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 19.164 ms | 100 - 100 MB | NPU | [Swin-Small.onnx](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.onnx) |
67
- | Swin-Small | w8a16 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | ONNX | 103.531 ms | 280 - 436 MB | NPU | [Swin-Small.onnx](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small_w8a16.onnx) |
68
- | Swin-Small | w8a16 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 79.335 ms | 282 - 525 MB | NPU | [Swin-Small.onnx](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small_w8a16.onnx) |
69
- | Swin-Small | w8a16 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 77.508 ms | 288 - 494 MB | NPU | [Swin-Small.onnx](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small_w8a16.onnx) |
70
- | Swin-Small | w8a16 | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 103.142 ms | 463 - 463 MB | NPU | [Swin-Small.onnx](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small_w8a16.onnx) |
71
 
72
 
73
 
@@ -131,8 +130,8 @@ Profiling Results
131
  Swin-Small
132
  Device : cs_8275 (ANDROID 14)
133
  Runtime : TFLITE
134
- Estimated inference time (ms) : 44.5
135
- Estimated peak memory usage (MB): [0, 258]
136
  Total # Ops : 1563
137
  Compute Unit(s) : npu (1563 ops) gpu (0 ops) cpu (0 ops)
138
  ```
@@ -221,13 +220,13 @@ AI Hub. [Sign up for access](https://myaccount.qualcomm.com/signup).
221
  You can also run the demo on-device.
222
 
223
  ```bash
224
- python -m qai_hub_models.models.swin_small.demo --on-device
225
  ```
226
 
227
  **NOTE**: If you want running in a Jupyter Notebook or Google Colab like
228
  environment, please add the following to your cell (instead of the above).
229
  ```
230
- %run -m qai_hub_models.models.swin_small.demo -- --on-device
231
  ```
232
 
233
 
 
35
 
36
  | Model | Precision | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit | Target Model
37
  |---|---|---|---|---|---|---|---|---|
38
+ | Swin-Small | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 177.526 ms | 0 - 278 MB | NPU | [Swin-Small.tflite](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.tflite) |
39
+ | Swin-Small | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN | 166.796 ms | 1 - 10 MB | NPU | Use Export Script |
40
+ | Swin-Small | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 23.945 ms | 0 - 271 MB | NPU | [Swin-Small.tflite](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.tflite) |
41
+ | Swin-Small | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN | 24.645 ms | 1 - 243 MB | NPU | Use Export Script |
42
+ | Swin-Small | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 18.868 ms | 0 - 28 MB | NPU | [Swin-Small.tflite](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.tflite) |
43
+ | Swin-Small | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN | 16.438 ms | 1 - 3 MB | NPU | Use Export Script |
44
+ | Swin-Small | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 20.998 ms | 0 - 278 MB | NPU | [Swin-Small.tflite](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.tflite) |
45
+ | Swin-Small | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN | 18.428 ms | 1 - 10 MB | NPU | Use Export Script |
46
+ | Swin-Small | float | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 177.526 ms | 0 - 278 MB | NPU | [Swin-Small.tflite](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.tflite) |
47
+ | Swin-Small | float | SA7255P ADP | Qualcomm® SA7255P | QNN | 166.796 ms | 1 - 10 MB | NPU | Use Export Script |
48
+ | Swin-Small | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 18.908 ms | 0 - 31 MB | NPU | [Swin-Small.tflite](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.tflite) |
49
+ | Swin-Small | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN | 16.611 ms | 2 - 4 MB | NPU | Use Export Script |
50
+ | Swin-Small | float | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 26.708 ms | 0 - 271 MB | NPU | [Swin-Small.tflite](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.tflite) |
51
+ | Swin-Small | float | SA8295P ADP | Qualcomm® SA8295P | QNN | 23.773 ms | 1 - 17 MB | NPU | Use Export Script |
52
+ | Swin-Small | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 18.958 ms | 0 - 26 MB | NPU | [Swin-Small.tflite](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.tflite) |
53
+ | Swin-Small | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN | 16.592 ms | 1 - 2 MB | NPU | Use Export Script |
54
+ | Swin-Small | float | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 20.998 ms | 0 - 278 MB | NPU | [Swin-Small.tflite](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.tflite) |
55
+ | Swin-Small | float | SA8775P ADP | Qualcomm® SA8775P | QNN | 18.428 ms | 1 - 10 MB | NPU | Use Export Script |
56
+ | Swin-Small | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | TFLITE | 18.971 ms | 0 - 29 MB | NPU | [Swin-Small.tflite](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.tflite) |
57
+ | Swin-Small | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | QNN | 16.567 ms | 0 - 58 MB | NPU | Use Export Script |
58
+ | Swin-Small | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | ONNX | 16.211 ms | 0 - 282 MB | NPU | [Swin-Small.onnx](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.onnx) |
59
+ | Swin-Small | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 12.778 ms | 0 - 280 MB | NPU | [Swin-Small.tflite](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.tflite) |
60
+ | Swin-Small | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN | 10.804 ms | 1 - 768 MB | NPU | Use Export Script |
61
+ | Swin-Small | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 10.672 ms | 1 - 746 MB | NPU | [Swin-Small.onnx](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.onnx) |
62
+ | Swin-Small | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | QNN | 9.751 ms | 1 - 250 MB | NPU | Use Export Script |
63
+ | Swin-Small | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 9.627 ms | 1 - 517 MB | NPU | [Swin-Small.onnx](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.onnx) |
64
+ | Swin-Small | float | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN | 17.506 ms | 1 - 1 MB | NPU | Use Export Script |
65
+ | Swin-Small | float | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 18.486 ms | 100 - 100 MB | NPU | [Swin-Small.onnx](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.onnx) |
66
+ | Swin-Small | w8a16 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | ONNX | 94.708 ms | 271 - 429 MB | NPU | [Swin-Small.onnx](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small_w8a16.onnx) |
67
+ | Swin-Small | w8a16 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 77.878 ms | 286 - 555 MB | NPU | [Swin-Small.onnx](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small_w8a16.onnx) |
68
+ | Swin-Small | w8a16 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 63.95 ms | 285 - 518 MB | NPU | [Swin-Small.onnx](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small_w8a16.onnx) |
69
+ | Swin-Small | w8a16 | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 99.309 ms | 463 - 463 MB | NPU | [Swin-Small.onnx](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small_w8a16.onnx) |
 
70
 
71
 
72
 
 
130
  Swin-Small
131
  Device : cs_8275 (ANDROID 14)
132
  Runtime : TFLITE
133
+ Estimated inference time (ms) : 177.5
134
+ Estimated peak memory usage (MB): [0, 278]
135
  Total # Ops : 1563
136
  Compute Unit(s) : npu (1563 ops) gpu (0 ops) cpu (0 ops)
137
  ```
 
220
  You can also run the demo on-device.
221
 
222
  ```bash
223
+ python -m qai_hub_models.models.swin_small.demo --eval-mode on-device
224
  ```
225
 
226
  **NOTE**: If you want running in a Jupyter Notebook or Google Colab like
227
  environment, please add the following to your cell (instead of the above).
228
  ```
229
+ %run -m qai_hub_models.models.swin_small.demo -- --eval-mode on-device
230
  ```
231
 
232
 
Swin-Small.onnx CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:7d9b8b50450a9be45503ff2f03f7830bd10315f8c75302420050278fce685741
3
- size 202036385
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:57176a7e7fe80562ba3666ad431d648bc38fc5c9c76be66f977914ccfee0f3a3
3
+ size 202036446
Swin-Small_w8a16.onnx CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:8e8b009624042e52f318519507bdc618001afbc6d31fd56a9046cc83be9bd09d
3
- size 202937683
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:60d28b12d664fa7f7a1cbeab7dc76d84e47849532d4bd41c91b2949b3c0062ad
3
+ size 203018194