Hugging Face's logo

qualcomm
/
Swin-Tiny

Image Classification
PyTorch
LiteRT
backbone
bu_auto
android
Model card Files
xet
Community
qaihm-bot commited on
Commit
f010f05
·
verified ·
1 Parent(s): 5fc60f1

v0.30.5

Browse files

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

Files changed (3) hide show
  1. README.md +37 -37
  2. Swin-Tiny.onnx +2 -2
  3. Swin-Tiny_w8a16.onnx +2 -2
README.md CHANGED
@@ -35,39 +35,39 @@ 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-Tiny | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 24.932 ms | 0 - 153 MB | NPU | [Swin-Tiny.tflite](https://huggingface.co/qualcomm/Swin-Tiny/blob/main/Swin-Tiny.tflite) |
39
- | Swin-Tiny | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN | 24.472 ms | 1 - 10 MB | NPU | Use Export Script |
40
- | Swin-Tiny | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 13.88 ms | 0 - 157 MB | NPU | [Swin-Tiny.tflite](https://huggingface.co/qualcomm/Swin-Tiny/blob/main/Swin-Tiny.tflite) |
41
- | Swin-Tiny | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN | 16.357 ms | 1 - 153 MB | NPU | Use Export Script |
42
- | Swin-Tiny | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 10.994 ms | 0 - 16 MB | NPU | [Swin-Tiny.tflite](https://huggingface.co/qualcomm/Swin-Tiny/blob/main/Swin-Tiny.tflite) |
43
- | Swin-Tiny | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN | 11.012 ms | 1 - 3 MB | NPU | Use Export Script |
44
- | Swin-Tiny | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 12.339 ms | 0 - 153 MB | NPU | [Swin-Tiny.tflite](https://huggingface.co/qualcomm/Swin-Tiny/blob/main/Swin-Tiny.tflite) |
45
- | Swin-Tiny | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN | 12.184 ms | 1 - 13 MB | NPU | Use Export Script |
46
- | Swin-Tiny | float | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 24.932 ms | 0 - 153 MB | NPU | [Swin-Tiny.tflite](https://huggingface.co/qualcomm/Swin-Tiny/blob/main/Swin-Tiny.tflite) |
47
- | Swin-Tiny | float | SA7255P ADP | Qualcomm® SA7255P | QNN | 24.472 ms | 1 - 10 MB | NPU | Use Export Script |
48
- | Swin-Tiny | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 11.031 ms | 0 - 15 MB | NPU | [Swin-Tiny.tflite](https://huggingface.co/qualcomm/Swin-Tiny/blob/main/Swin-Tiny.tflite) |
49
- | Swin-Tiny | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN | 11.018 ms | 1 - 5 MB | NPU | Use Export Script |
50
- | Swin-Tiny | float | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 15.29 ms | 0 - 151 MB | NPU | [Swin-Tiny.tflite](https://huggingface.co/qualcomm/Swin-Tiny/blob/main/Swin-Tiny.tflite) |
51
- | Swin-Tiny | float | SA8295P ADP | Qualcomm® SA8295P | QNN | 15.141 ms | 1 - 18 MB | NPU | Use Export Script |
52
- | Swin-Tiny | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 11.096 ms | 0 - 18 MB | NPU | [Swin-Tiny.tflite](https://huggingface.co/qualcomm/Swin-Tiny/blob/main/Swin-Tiny.tflite) |
53
- | Swin-Tiny | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN | 11.093 ms | 1 - 3 MB | NPU | Use Export Script |
54
- | Swin-Tiny | float | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 12.339 ms | 0 - 153 MB | NPU | [Swin-Tiny.tflite](https://huggingface.co/qualcomm/Swin-Tiny/blob/main/Swin-Tiny.tflite) |
55
- | Swin-Tiny | float | SA8775P ADP | Qualcomm® SA8775P | QNN | 12.184 ms | 1 - 13 MB | NPU | Use Export Script |
56
- | Swin-Tiny | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | TFLITE | 11.024 ms | 0 - 15 MB | NPU | [Swin-Tiny.tflite](https://huggingface.co/qualcomm/Swin-Tiny/blob/main/Swin-Tiny.tflite) |
57
- | Swin-Tiny | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | QNN | 10.964 ms | 0 - 36 MB | NPU | Use Export Script |
58
- | Swin-Tiny | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | ONNX | 9.693 ms | 0 - 191 MB | NPU | [Swin-Tiny.onnx](https://huggingface.co/qualcomm/Swin-Tiny/blob/main/Swin-Tiny.onnx) |
59
- | Swin-Tiny | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 7.395 ms | 0 - 161 MB | NPU | [Swin-Tiny.tflite](https://huggingface.co/qualcomm/Swin-Tiny/blob/main/Swin-Tiny.tflite) |
60
- | Swin-Tiny | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN | 7.403 ms | 1 - 162 MB | NPU | Use Export Script |
61
- | Swin-Tiny | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 6.489 ms | 1 - 158 MB | NPU | [Swin-Tiny.onnx](https://huggingface.co/qualcomm/Swin-Tiny/blob/main/Swin-Tiny.onnx) |
62
- | Swin-Tiny | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | TFLITE | 5.648 ms | 0 - 154 MB | NPU | [Swin-Tiny.tflite](https://huggingface.co/qualcomm/Swin-Tiny/blob/main/Swin-Tiny.tflite) |
63
- | Swin-Tiny | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | QNN | 6.707 ms | 1 - 151 MB | NPU | Use Export Script |
64
- | Swin-Tiny | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 6.04 ms | 1 - 148 MB | NPU | [Swin-Tiny.onnx](https://huggingface.co/qualcomm/Swin-Tiny/blob/main/Swin-Tiny.onnx) |
65
- | Swin-Tiny | float | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN | 11.406 ms | 1 - 1 MB | NPU | Use Export Script |
66
- | Swin-Tiny | float | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 11.335 ms | 57 - 57 MB | NPU | [Swin-Tiny.onnx](https://huggingface.co/qualcomm/Swin-Tiny/blob/main/Swin-Tiny.onnx) |
67
- | Swin-Tiny | w8a16 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | ONNX | 60.356 ms | 149 - 258 MB | NPU | [Swin-Tiny.onnx](https://huggingface.co/qualcomm/Swin-Tiny/blob/main/Swin-Tiny_w8a16.onnx) |
68
- | Swin-Tiny | w8a16 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 48.558 ms | 157 - 315 MB | NPU | [Swin-Tiny.onnx](https://huggingface.co/qualcomm/Swin-Tiny/blob/main/Swin-Tiny_w8a16.onnx) |
69
- | Swin-Tiny | w8a16 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 43.226 ms | 158 - 290 MB | NPU | [Swin-Tiny.onnx](https://huggingface.co/qualcomm/Swin-Tiny/blob/main/Swin-Tiny_w8a16.onnx) |
70
- | Swin-Tiny | w8a16 | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 62.702 ms | 232 - 232 MB | NPU | [Swin-Tiny.onnx](https://huggingface.co/qualcomm/Swin-Tiny/blob/main/Swin-Tiny_w8a16.onnx) |
71
 
72
 
73
 
@@ -131,8 +131,8 @@ Profiling Results
131
  Swin-Tiny
132
  Device : cs_8275 (ANDROID 14)
133
  Runtime : TFLITE
134
- Estimated inference time (ms) : 24.9
135
- Estimated peak memory usage (MB): [0, 153]
136
  Total # Ops : 837
137
  Compute Unit(s) : npu (837 ops) gpu (0 ops) cpu (0 ops)
138
  ```
@@ -221,13 +221,13 @@ 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_tiny.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_tiny.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-Tiny | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 88.309 ms | 0 - 173 MB | NPU | [Swin-Tiny.tflite](https://huggingface.co/qualcomm/Swin-Tiny/blob/main/Swin-Tiny.tflite) |
39
+ | Swin-Tiny | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN | 83.759 ms | 1 - 10 MB | NPU | Use Export Script |
40
+ | Swin-Tiny | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 13.875 ms | 0 - 170 MB | NPU | [Swin-Tiny.tflite](https://huggingface.co/qualcomm/Swin-Tiny/blob/main/Swin-Tiny.tflite) |
41
+ | Swin-Tiny | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN | 14.849 ms | 1 - 155 MB | NPU | Use Export Script |
42
+ | Swin-Tiny | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 11.062 ms | 0 - 22 MB | NPU | [Swin-Tiny.tflite](https://huggingface.co/qualcomm/Swin-Tiny/blob/main/Swin-Tiny.tflite) |
43
+ | Swin-Tiny | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN | 9.907 ms | 1 - 4 MB | NPU | Use Export Script |
44
+ | Swin-Tiny | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 12.332 ms | 8 - 182 MB | NPU | [Swin-Tiny.tflite](https://huggingface.co/qualcomm/Swin-Tiny/blob/main/Swin-Tiny.tflite) |
45
+ | Swin-Tiny | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN | 11.024 ms | 1 - 11 MB | NPU | Use Export Script |
46
+ | Swin-Tiny | float | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 88.309 ms | 0 - 173 MB | NPU | [Swin-Tiny.tflite](https://huggingface.co/qualcomm/Swin-Tiny/blob/main/Swin-Tiny.tflite) |
47
+ | Swin-Tiny | float | SA7255P ADP | Qualcomm® SA7255P | QNN | 83.759 ms | 1 - 10 MB | NPU | Use Export Script |
48
+ | Swin-Tiny | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 11.012 ms | 0 - 26 MB | NPU | [Swin-Tiny.tflite](https://huggingface.co/qualcomm/Swin-Tiny/blob/main/Swin-Tiny.tflite) |
49
+ | Swin-Tiny | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN | 9.881 ms | 1 - 3 MB | NPU | Use Export Script |
50
+ | Swin-Tiny | float | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 15.902 ms | 0 - 172 MB | NPU | [Swin-Tiny.tflite](https://huggingface.co/qualcomm/Swin-Tiny/blob/main/Swin-Tiny.tflite) |
51
+ | Swin-Tiny | float | SA8295P ADP | Qualcomm® SA8295P | QNN | 14.319 ms | 1 - 18 MB | NPU | Use Export Script |
52
+ | Swin-Tiny | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 11.099 ms | 0 - 23 MB | NPU | [Swin-Tiny.tflite](https://huggingface.co/qualcomm/Swin-Tiny/blob/main/Swin-Tiny.tflite) |
53
+ | Swin-Tiny | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN | 9.905 ms | 1 - 3 MB | NPU | Use Export Script |
54
+ | Swin-Tiny | float | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 12.332 ms | 8 - 182 MB | NPU | [Swin-Tiny.tflite](https://huggingface.co/qualcomm/Swin-Tiny/blob/main/Swin-Tiny.tflite) |
55
+ | Swin-Tiny | float | SA8775P ADP | Qualcomm® SA8775P | QNN | 11.024 ms | 1 - 11 MB | NPU | Use Export Script |
56
+ | Swin-Tiny | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | TFLITE | 11.029 ms | 0 - 21 MB | NPU | [Swin-Tiny.tflite](https://huggingface.co/qualcomm/Swin-Tiny/blob/main/Swin-Tiny.tflite) |
57
+ | Swin-Tiny | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | QNN | 9.819 ms | 0 - 38 MB | NPU | Use Export Script |
58
+ | Swin-Tiny | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | ONNX | 9.324 ms | 0 - 199 MB | NPU | [Swin-Tiny.onnx](https://huggingface.co/qualcomm/Swin-Tiny/blob/main/Swin-Tiny.onnx) |
59
+ | Swin-Tiny | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 7.261 ms | 0 - 180 MB | NPU | [Swin-Tiny.tflite](https://huggingface.co/qualcomm/Swin-Tiny/blob/main/Swin-Tiny.tflite) |
60
+ | Swin-Tiny | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN | 6.321 ms | 1 - 168 MB | NPU | Use Export Script |
61
+ | Swin-Tiny | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 5.991 ms | 0 - 170 MB | NPU | [Swin-Tiny.onnx](https://huggingface.co/qualcomm/Swin-Tiny/blob/main/Swin-Tiny.onnx) |
62
+ | Swin-Tiny | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | TFLITE | 7.098 ms | 0 - 170 MB | NPU | [Swin-Tiny.tflite](https://huggingface.co/qualcomm/Swin-Tiny/blob/main/Swin-Tiny.tflite) |
63
+ | Swin-Tiny | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | QNN | 4.844 ms | 1 - 157 MB | NPU | Use Export Script |
64
+ | Swin-Tiny | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 5.717 ms | 1 - 158 MB | NPU | [Swin-Tiny.onnx](https://huggingface.co/qualcomm/Swin-Tiny/blob/main/Swin-Tiny.onnx) |
65
+ | Swin-Tiny | float | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN | 10.537 ms | 1 - 1 MB | NPU | Use Export Script |
66
+ | Swin-Tiny | float | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 11.015 ms | 57 - 57 MB | NPU | [Swin-Tiny.onnx](https://huggingface.co/qualcomm/Swin-Tiny/blob/main/Swin-Tiny.onnx) |
67
+ | Swin-Tiny | w8a16 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | ONNX | 57.363 ms | 124 - 244 MB | NPU | [Swin-Tiny.onnx](https://huggingface.co/qualcomm/Swin-Tiny/blob/main/Swin-Tiny_w8a16.onnx) |
68
+ | Swin-Tiny | w8a16 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 46.75 ms | 156 - 325 MB | NPU | [Swin-Tiny.onnx](https://huggingface.co/qualcomm/Swin-Tiny/blob/main/Swin-Tiny_w8a16.onnx) |
69
+ | Swin-Tiny | w8a16 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 37.373 ms | 157 - 302 MB | NPU | [Swin-Tiny.onnx](https://huggingface.co/qualcomm/Swin-Tiny/blob/main/Swin-Tiny_w8a16.onnx) |
70
+ | Swin-Tiny | w8a16 | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 53.469 ms | 230 - 230 MB | NPU | [Swin-Tiny.onnx](https://huggingface.co/qualcomm/Swin-Tiny/blob/main/Swin-Tiny_w8a16.onnx) |
71
 
72
 
73
 
 
131
  Swin-Tiny
132
  Device : cs_8275 (ANDROID 14)
133
  Runtime : TFLITE
134
+ Estimated inference time (ms) : 88.3
135
+ Estimated peak memory usage (MB): [0, 173]
136
  Total # Ops : 837
137
  Compute Unit(s) : npu (837 ops) gpu (0 ops) cpu (0 ops)
138
  ```
 
221
  You can also run the demo on-device.
222
 
223
  ```bash
224
+ python -m qai_hub_models.models.swin_tiny.demo --eval-mode 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_tiny.demo -- --eval-mode on-device
231
  ```
232
 
233
 
Swin-Tiny.onnx CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:8907e08229fad5704fe18abcb567ea163cec23de0ff32e5ea368752e7e13784c
3
- size 115347412
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:af276e1b9b9a5cbf1aefe200e1e54efab3a0f241877065779ae007b891549d6c
3
+ size 115347473
Swin-Tiny_w8a16.onnx CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:6ee446b9e7b8aab8a5ab511a2e1fbf78ab489488f346936516b1874b126aa4a0
3
- size 115849045
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3994bda76bec730ba5e3735cb69892d5b1f597150f9ad45831c0fa7278b48cdc
3
+ size 115894555