Hugging Face's logo

qualcomm
/
EfficientNet-B4

Image Classification
PyTorch
LiteRT
backbone
bu_auto
android
Model card Files
xet
Community
qaihm-bot commited on
Commit
70d3d84
·
verified ·
1 Parent(s): 6aa63ce

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +41 -28
README.md CHANGED
@@ -1,6 +1,6 @@
1
  ---
2
  library_name: pytorch
3
- license: bsd-3-clause
4
  tags:
5
  - backbone
6
  - android
@@ -26,34 +26,47 @@ More details on model performance across various devices, can be found
26
 
27
  ### Model Details
28
 
29
- - **Model Type:** Image classification
30
  - **Model Stats:**
31
  - Model checkpoint: Imagenet
32
  - Input resolution: 380x380
33
  - Number of parameters: 19.34M
34
  - Model size: 74.5 MB
35
 
36
- | Model | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Precision | Primary Compute Unit | Target Model
37
  |---|---|---|---|---|---|---|---|---|
38
- | EfficientNet-B4 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | TFLITE | 3.305 ms | 0 - 383 MB | FP16 | NPU | [EfficientNet-B4.tflite](https://huggingface.co/qualcomm/EfficientNet-B4/blob/main/EfficientNet-B4.tflite) |
39
- | EfficientNet-B4 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | QNN | 3.337 ms | 0 - 217 MB | FP16 | NPU | [EfficientNet-B4.so](https://huggingface.co/qualcomm/EfficientNet-B4/blob/main/EfficientNet-B4.so) |
40
- | EfficientNet-B4 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | ONNX | 3.203 ms | 0 - 102 MB | FP16 | NPU | [EfficientNet-B4.onnx](https://huggingface.co/qualcomm/EfficientNet-B4/blob/main/EfficientNet-B4.onnx) |
41
- | EfficientNet-B4 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | TFLITE | 2.433 ms | 0 - 77 MB | FP16 | NPU | [EfficientNet-B4.tflite](https://huggingface.co/qualcomm/EfficientNet-B4/blob/main/EfficientNet-B4.tflite) |
42
- | EfficientNet-B4 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | QNN | 2.45 ms | 1 - 47 MB | FP16 | NPU | [EfficientNet-B4.so](https://huggingface.co/qualcomm/EfficientNet-B4/blob/main/EfficientNet-B4.so) |
43
- | EfficientNet-B4 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | ONNX | 2.407 ms | 0 - 58 MB | FP16 | NPU | [EfficientNet-B4.onnx](https://huggingface.co/qualcomm/EfficientNet-B4/blob/main/EfficientNet-B4.onnx) |
44
- | EfficientNet-B4 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | TFLITE | 1.938 ms | 0 - 65 MB | FP16 | NPU | [EfficientNet-B4.tflite](https://huggingface.co/qualcomm/EfficientNet-B4/blob/main/EfficientNet-B4.tflite) |
45
- | EfficientNet-B4 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | QNN | 1.945 ms | 1 - 32 MB | FP16 | NPU | Use Export Script |
46
- | EfficientNet-B4 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | ONNX | 2.292 ms | 0 - 37 MB | FP16 | NPU | [EfficientNet-B4.onnx](https://huggingface.co/qualcomm/EfficientNet-B4/blob/main/EfficientNet-B4.onnx) |
47
- | EfficientNet-B4 | QCS8275 (Proxy) | QCS8275 Proxy | TFLITE | 64.813 ms | 0 - 61 MB | FP16 | NPU | [EfficientNet-B4.tflite](https://huggingface.co/qualcomm/EfficientNet-B4/blob/main/EfficientNet-B4.tflite) |
48
- | EfficientNet-B4 | QCS8275 (Proxy) | QCS8275 Proxy | QNN | 64.527 ms | 1 - 10 MB | FP16 | NPU | Use Export Script |
49
- | EfficientNet-B4 | QCS8550 (Proxy) | QCS8550 Proxy | TFLITE | 3.32 ms | 0 - 378 MB | FP16 | NPU | [EfficientNet-B4.tflite](https://huggingface.co/qualcomm/EfficientNet-B4/blob/main/EfficientNet-B4.tflite) |
50
- | EfficientNet-B4 | QCS8550 (Proxy) | QCS8550 Proxy | QNN | 3.171 ms | 1 - 3 MB | FP16 | NPU | Use Export Script |
51
- | EfficientNet-B4 | QCS9075 (Proxy) | QCS9075 Proxy | TFLITE | 5.264 ms | 0 - 61 MB | FP16 | NPU | [EfficientNet-B4.tflite](https://huggingface.co/qualcomm/EfficientNet-B4/blob/main/EfficientNet-B4.tflite) |
52
- | EfficientNet-B4 | QCS9075 (Proxy) | QCS9075 Proxy | QNN | 5.043 ms | 1 - 10 MB | FP16 | NPU | Use Export Script |
53
- | EfficientNet-B4 | QCS8450 (Proxy) | QCS8450 Proxy | TFLITE | 7.198 ms | 0 - 80 MB | FP16 | NPU | [EfficientNet-B4.tflite](https://huggingface.co/qualcomm/EfficientNet-B4/blob/main/EfficientNet-B4.tflite) |
54
- | EfficientNet-B4 | QCS8450 (Proxy) | QCS8450 Proxy | QNN | 7.254 ms | 1 - 46 MB | FP16 | NPU | Use Export Script |
55
- | EfficientNet-B4 | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN | 3.459 ms | 1 - 1 MB | FP16 | NPU | Use Export Script |
56
- | EfficientNet-B4 | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 3.451 ms | 45 - 45 MB | FP16 | NPU | [EfficientNet-B4.onnx](https://huggingface.co/qualcomm/EfficientNet-B4/blob/main/EfficientNet-B4.onnx) |
 
 
 
 
 
 
 
 
 
 
 
 
 
57
 
58
 
59
 
@@ -115,12 +128,12 @@ python -m qai_hub_models.models.efficientnet_b4.export
115
  Profiling Results
116
  ------------------------------------------------------------
117
  EfficientNet-B4
118
- Device : Samsung Galaxy S23 (13)
119
- Runtime : TFLITE
120
- Estimated inference time (ms) : 3.3
121
- Estimated peak memory usage (MB): [0, 383]
122
- Total # Ops : 482
123
- Compute Unit(s) : NPU (482 ops)
124
  ```
125
 
126
 
 
1
  ---
2
  library_name: pytorch
3
+ license: other
4
  tags:
5
  - backbone
6
  - android
 
26
 
27
  ### Model Details
28
 
29
+ - **Model Type:** Model_use_case.image_classification
30
  - **Model Stats:**
31
  - Model checkpoint: Imagenet
32
  - Input resolution: 380x380
33
  - Number of parameters: 19.34M
34
  - Model size: 74.5 MB
35
 
36
+ | Model | Precision | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit | Target Model
37
  |---|---|---|---|---|---|---|---|---|
38
+ | EfficientNet-B4 | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 12.273 ms | 0 - 61 MB | NPU | [EfficientNet-B4.tflite](https://huggingface.co/qualcomm/EfficientNet-B4/blob/main/EfficientNet-B4.tflite) |
39
+ | EfficientNet-B4 | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN | 11.919 ms | 1 - 10 MB | NPU | Use Export Script |
40
+ | EfficientNet-B4 | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 7.224 ms | 0 - 79 MB | NPU | [EfficientNet-B4.tflite](https://huggingface.co/qualcomm/EfficientNet-B4/blob/main/EfficientNet-B4.tflite) |
41
+ | EfficientNet-B4 | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN | 7.981 ms | 1 - 44 MB | NPU | Use Export Script |
42
+ | EfficientNet-B4 | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 3.327 ms | 0 - 385 MB | NPU | [EfficientNet-B4.tflite](https://huggingface.co/qualcomm/EfficientNet-B4/blob/main/EfficientNet-B4.tflite) |
43
+ | EfficientNet-B4 | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN | 3.186 ms | 1 - 3 MB | NPU | Use Export Script |
44
+ | EfficientNet-B4 | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 4.387 ms | 0 - 62 MB | NPU | [EfficientNet-B4.tflite](https://huggingface.co/qualcomm/EfficientNet-B4/blob/main/EfficientNet-B4.tflite) |
45
+ | EfficientNet-B4 | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN | 4.17 ms | 0 - 13 MB | NPU | Use Export Script |
46
+ | EfficientNet-B4 | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | TFLITE | 3.312 ms | 0 - 384 MB | NPU | [EfficientNet-B4.tflite](https://huggingface.co/qualcomm/EfficientNet-B4/blob/main/EfficientNet-B4.tflite) |
47
+ | EfficientNet-B4 | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | QNN | 3.189 ms | 0 - 29 MB | NPU | Use Export Script |
48
+ | EfficientNet-B4 | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | ONNX | 3.175 ms | 0 - 152 MB | NPU | [EfficientNet-B4.onnx](https://huggingface.co/qualcomm/EfficientNet-B4/blob/main/EfficientNet-B4.onnx) |
49
+ | EfficientNet-B4 | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 2.453 ms | 0 - 79 MB | NPU | [EfficientNet-B4.tflite](https://huggingface.co/qualcomm/EfficientNet-B4/blob/main/EfficientNet-B4.tflite) |
50
+ | EfficientNet-B4 | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN | 2.356 ms | 1 - 47 MB | NPU | Use Export Script |
51
+ | EfficientNet-B4 | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 2.364 ms | 0 - 55 MB | NPU | [EfficientNet-B4.onnx](https://huggingface.co/qualcomm/EfficientNet-B4/blob/main/EfficientNet-B4.onnx) |
52
+ | EfficientNet-B4 | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | TFLITE | 2.281 ms | 0 - 64 MB | NPU | [EfficientNet-B4.tflite](https://huggingface.co/qualcomm/EfficientNet-B4/blob/main/EfficientNet-B4.tflite) |
53
+ | EfficientNet-B4 | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | QNN | 2.168 ms | 1 - 35 MB | NPU | Use Export Script |
54
+ | EfficientNet-B4 | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 2.272 ms | 0 - 36 MB | NPU | [EfficientNet-B4.onnx](https://huggingface.co/qualcomm/EfficientNet-B4/blob/main/EfficientNet-B4.onnx) |
55
+ | EfficientNet-B4 | float | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN | 3.458 ms | 1 - 1 MB | NPU | Use Export Script |
56
+ | EfficientNet-B4 | float | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 3.47 ms | 46 - 46 MB | NPU | [EfficientNet-B4.onnx](https://huggingface.co/qualcomm/EfficientNet-B4/blob/main/EfficientNet-B4.onnx) |
57
+ | EfficientNet-B4 | w8a16 | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN | 6.535 ms | 0 - 10 MB | NPU | Use Export Script |
58
+ | EfficientNet-B4 | w8a16 | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN | 4.378 ms | 0 - 62 MB | NPU | Use Export Script |
59
+ | EfficientNet-B4 | w8a16 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN | 3.423 ms | 0 - 4 MB | NPU | Use Export Script |
60
+ | EfficientNet-B4 | w8a16 | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN | 3.851 ms | 0 - 15 MB | NPU | Use Export Script |
61
+ | EfficientNet-B4 | w8a16 | RB3 Gen 2 (Proxy) | Qualcomm® QCS6490 (Proxy) | QNN | 11.744 ms | 0 - 14 MB | NPU | Use Export Script |
62
+ | EfficientNet-B4 | w8a16 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | QNN | 3.451 ms | 0 - 16 MB | NPU | Use Export Script |
63
+ | EfficientNet-B4 | w8a16 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | ONNX | 3.663 ms | 0 - 36 MB | NPU | [EfficientNet-B4.onnx](https://huggingface.co/qualcomm/EfficientNet-B4/blob/main/EfficientNet-B4_w8a16.onnx) |
64
+ | EfficientNet-B4 | w8a16 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN | 2.301 ms | 0 - 69 MB | NPU | Use Export Script |
65
+ | EfficientNet-B4 | w8a16 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 2.451 ms | 0 - 90 MB | NPU | [EfficientNet-B4.onnx](https://huggingface.co/qualcomm/EfficientNet-B4/blob/main/EfficientNet-B4_w8a16.onnx) |
66
+ | EfficientNet-B4 | w8a16 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | QNN | 1.936 ms | 0 - 53 MB | NPU | Use Export Script |
67
+ | EfficientNet-B4 | w8a16 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 2.236 ms | 0 - 68 MB | NPU | [EfficientNet-B4.onnx](https://huggingface.co/qualcomm/EfficientNet-B4/blob/main/EfficientNet-B4_w8a16.onnx) |
68
+ | EfficientNet-B4 | w8a16 | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN | 3.764 ms | 0 - 0 MB | NPU | Use Export Script |
69
+ | EfficientNet-B4 | w8a16 | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 3.85 ms | 26 - 26 MB | NPU | [EfficientNet-B4.onnx](https://huggingface.co/qualcomm/EfficientNet-B4/blob/main/EfficientNet-B4_w8a16.onnx) |
70
 
71
 
72
 
 
128
  Profiling Results
129
  ------------------------------------------------------------
130
  EfficientNet-B4
131
+ Device : cs_8275 (ANDROID 14)
132
+ Runtime : TFLITE
133
+ Estimated inference time (ms) : 12.3
134
+ Estimated peak memory usage (MB): [0, 61]
135
+ Total # Ops : 482
136
+ Compute Unit(s) : npu (482 ops) gpu (0 ops) cpu (0 ops)
137
  ```
138
 
139