Hugging Face's logo

qualcomm
/
EfficientNet-B4

Image Classification
PyTorch
backbone
bu_auto
android
Model card Files
xet
 Community
qaihm-bot commited on
Commit
5795edb
·
verified ·
1 Parent(s): 9570927

v0.48.0

Browse files

See https://github.com/qualcomm/ai-hub-models/releases/v0.48.0 for changelog.

Files changed (1) hide show
  1. README.md +43 -43
README.md CHANGED
@@ -16,7 +16,7 @@ pipeline_tag: image-classification
16
  EfficientNetB4 is a machine learning model that can classify images from the Imagenet dataset. It can also be used as a backbone in building more complex models for specific use cases.
17
 
18
  This is based on the implementation of EfficientNet-B4 found [here](https://github.com/pytorch/vision/blob/main/torchvision/models/efficientnet.py).
19
- This repository contains pre-exported model files optimized for Qualcomm® devices. You can use the [Qualcomm® AI Hub Models](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/models/efficientnet_b4) library to export with custom configurations. More details on model performance across various devices, can be found [here](#performance-summary).
20
 
21
  Qualcomm AI Hub Models uses [Qualcomm AI Hub Workbench](https://workbench.aihub.qualcomm.com) to compile, profile, and evaluate this model. [Sign up](https://myaccount.qualcomm.com/signup) to run these models on a hosted Qualcomm® device.
22
 
@@ -29,25 +29,25 @@ Below are pre-exported model assets ready for deployment.
29
 
30
  | Runtime | Precision | Chipset | SDK Versions | Download |
31
  |---|---|---|---|---|
32
- | ONNX | float | Universal | QAIRT 2.42, ONNX Runtime 1.24.1 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/efficientnet_b4/releases/v0.47.0/efficientnet_b4-onnx-float.zip)
33
- | ONNX | w8a16 | Universal | QAIRT 2.42, ONNX Runtime 1.24.1 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/efficientnet_b4/releases/v0.47.0/efficientnet_b4-onnx-w8a16.zip)
34
- | QNN_DLC | float | Universal | QAIRT 2.43 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/efficientnet_b4/releases/v0.47.0/efficientnet_b4-qnn_dlc-float.zip)
35
- | QNN_DLC | w8a16 | Universal | QAIRT 2.43 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/efficientnet_b4/releases/v0.47.0/efficientnet_b4-qnn_dlc-w8a16.zip)
36
- | TFLITE | float | Universal | QAIRT 2.43, TFLite 2.17.0 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/efficientnet_b4/releases/v0.47.0/efficientnet_b4-tflite-float.zip)
37
 
38
  For more device-specific assets and performance metrics, visit **[EfficientNet-B4 on Qualcomm® AI Hub](https://aihub.qualcomm.com/models/efficientnet_b4)**.
39
 
40
 
41
  ### Option 2: Export with Custom Configurations
42
 
43
- Use the [Qualcomm® AI Hub Models](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/models/efficientnet_b4) Python library to compile and export the model with your own:
44
  - Custom weights (e.g., fine-tuned checkpoints)
45
  - Custom input shapes
46
  - Target device and runtime configurations
47
 
48
  This option is ideal if you need to customize the model beyond the default configuration provided here.
49
 
50
- See our repository for [EfficientNet-B4 on GitHub](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/models/efficientnet_b4) for usage instructions.
51
 
52
  ## Model Details
53
 
@@ -63,41 +63,41 @@ See our repository for [EfficientNet-B4 on GitHub](https://github.com/quic/ai-hu
63
  ## Performance Summary
64
  | Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit
65
  |---|---|---|---|---|---|---
66
- | EfficientNet-B4 | ONNX | float | Snapdragon® X Elite | 3.351 ms | 45 - 45 MB | NPU
67
- | EfficientNet-B4 | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 2.249 ms | 0 - 127 MB | NPU
68
- | EfficientNet-B4 | ONNX | float | Qualcomm® QCS8550 (Proxy) | 3.049 ms | 0 - 52 MB | NPU
69
- | EfficientNet-B4 | ONNX | float | Qualcomm® QCS9075 | 4.017 ms | 0 - 4 MB | NPU
70
- | EfficientNet-B4 | ONNX | float | Snapdragon® 8 Elite For Galaxy Mobile | 1.763 ms | 0 - 77 MB | NPU
71
- | EfficientNet-B4 | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 1.471 ms | 0 - 77 MB | NPU
72
- | EfficientNet-B4 | ONNX | float | Snapdragon® X2 Elite | 1.632 ms | 45 - 45 MB | NPU
73
- | EfficientNet-B4 | QNN_DLC | float | Snapdragon® X Elite | 3.662 ms | 1 - 1 MB | NPU
74
- | EfficientNet-B4 | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 2.416 ms | 0 - 125 MB | NPU
75
- | EfficientNet-B4 | QNN_DLC | float | Qualcomm® QCS8275 (Proxy) | 12.016 ms | 1 - 69 MB | NPU
76
- | EfficientNet-B4 | QNN_DLC | float | Qualcomm® QCS8550 (Proxy) | 3.36 ms | 1 - 16 MB | NPU
77
- | EfficientNet-B4 | QNN_DLC | float | Qualcomm® QCS9075 | 4.199 ms | 3 - 5 MB | NPU
78
- | EfficientNet-B4 | QNN_DLC | float | Qualcomm® QCS8450 (Proxy) | 7.863 ms | 0 - 142 MB | NPU
79
- | EfficientNet-B4 | QNN_DLC | float | Snapdragon® 8 Elite For Galaxy Mobile | 1.862 ms | 0 - 74 MB | NPU
80
- | EfficientNet-B4 | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 Mobile | 1.51 ms | 1 - 75 MB | NPU
81
- | EfficientNet-B4 | QNN_DLC | float | Snapdragon® X2 Elite | 1.951 ms | 1 - 1 MB | NPU
82
- | EfficientNet-B4 | QNN_DLC | w8a16 | Snapdragon® X Elite | 3.794 ms | 0 - 0 MB | NPU
83
- | EfficientNet-B4 | QNN_DLC | w8a16 | Snapdragon® 8 Gen 3 Mobile | 2.305 ms | 0 - 154 MB | NPU
84
- | EfficientNet-B4 | QNN_DLC | w8a16 | Qualcomm® QCS6490 | 8.883 ms | 2 - 4 MB | NPU
85
- | EfficientNet-B4 | QNN_DLC | w8a16 | Qualcomm® QCS8275 (Proxy) | 6.65 ms | 0 - 99 MB | NPU
86
- | EfficientNet-B4 | QNN_DLC | w8a16 | Qualcomm® QCS8550 (Proxy) | 3.452 ms | 0 - 2 MB | NPU
87
- | EfficientNet-B4 | QNN_DLC | w8a16 | Qualcomm® QCS9075 | 3.8 ms | 0 - 2 MB | NPU
88
- | EfficientNet-B4 | QNN_DLC | w8a16 | Qualcomm® QCM6690 | 17.122 ms | 0 - 231 MB | NPU
89
- | EfficientNet-B4 | QNN_DLC | w8a16 | Qualcomm® QCS8450 (Proxy) | 4.106 ms | 0 - 154 MB | NPU
90
- | EfficientNet-B4 | QNN_DLC | w8a16 | Snapdragon® 8 Elite For Galaxy Mobile | 1.602 ms | 0 - 102 MB | NPU
91
- | EfficientNet-B4 | QNN_DLC | w8a16 | Snapdragon® 7 Gen 4 Mobile | 3.599 ms | 0 - 109 MB | NPU
92
- | EfficientNet-B4 | QNN_DLC | w8a16 | Snapdragon® 8 Elite Gen 5 Mobile | 1.323 ms | 0 - 102 MB | NPU
93
- | EfficientNet-B4 | QNN_DLC | w8a16 | Snapdragon® X2 Elite | 1.711 ms | 0 - 0 MB | NPU
94
- | EfficientNet-B4 | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 2.408 ms | 0 - 165 MB | NPU
95
- | EfficientNet-B4 | TFLITE | float | Qualcomm® QCS8275 (Proxy) | 11.994 ms | 0 - 105 MB | NPU
96
- | EfficientNet-B4 | TFLITE | float | Qualcomm® QCS8550 (Proxy) | 3.344 ms | 0 - 5 MB | NPU
97
- | EfficientNet-B4 | TFLITE | float | Qualcomm® QCS9075 | 4.195 ms | 0 - 48 MB | NPU
98
- | EfficientNet-B4 | TFLITE | float | Qualcomm® QCS8450 (Proxy) | 7.804 ms | 0 - 186 MB | NPU
99
- | EfficientNet-B4 | TFLITE | float | Snapdragon® 8 Elite For Galaxy Mobile | 1.854 ms | 0 - 110 MB | NPU
100
- | EfficientNet-B4 | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 1.511 ms | 0 - 106 MB | NPU
101
 
102
  ## License
103
  * The license for the original implementation of EfficientNet-B4 can be found
 
16
  EfficientNetB4 is a machine learning model that can classify images from the Imagenet dataset. It can also be used as a backbone in building more complex models for specific use cases.
17
 
18
  This is based on the implementation of EfficientNet-B4 found [here](https://github.com/pytorch/vision/blob/main/torchvision/models/efficientnet.py).
19
+ This repository contains pre-exported model files optimized for Qualcomm® devices. You can use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/main/qai_hub_models/models/efficientnet_b4) library to export with custom configurations. More details on model performance across various devices, can be found [here](#performance-summary).
20
 
21
  Qualcomm AI Hub Models uses [Qualcomm AI Hub Workbench](https://workbench.aihub.qualcomm.com) to compile, profile, and evaluate this model. [Sign up](https://myaccount.qualcomm.com/signup) to run these models on a hosted Qualcomm® device.
22
 
 
29
 
30
  | Runtime | Precision | Chipset | SDK Versions | Download |
31
  |---|---|---|---|---|
32
+ | ONNX | float | Universal | QAIRT 2.42, ONNX Runtime 1.24.1 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/efficientnet_b4/releases/v0.48.0/efficientnet_b4-onnx-float.zip)
33
+ | ONNX | w8a16 | Universal | QAIRT 2.42, ONNX Runtime 1.24.1 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/efficientnet_b4/releases/v0.48.0/efficientnet_b4-onnx-w8a16.zip)
34
+ | QNN_DLC | float | Universal | QAIRT 2.43 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/efficientnet_b4/releases/v0.48.0/efficientnet_b4-qnn_dlc-float.zip)
35
+ | QNN_DLC | w8a16 | Universal | QAIRT 2.43 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/efficientnet_b4/releases/v0.48.0/efficientnet_b4-qnn_dlc-w8a16.zip)
36
+ | TFLITE | float | Universal | QAIRT 2.43, TFLite 2.17.0 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/efficientnet_b4/releases/v0.48.0/efficientnet_b4-tflite-float.zip)
37
 
38
  For more device-specific assets and performance metrics, visit **[EfficientNet-B4 on Qualcomm® AI Hub](https://aihub.qualcomm.com/models/efficientnet_b4)**.
39
 
40
 
41
  ### Option 2: Export with Custom Configurations
42
 
43
+ Use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/main/qai_hub_models/models/efficientnet_b4) Python library to compile and export the model with your own:
44
  - Custom weights (e.g., fine-tuned checkpoints)
45
  - Custom input shapes
46
  - Target device and runtime configurations
47
 
48
  This option is ideal if you need to customize the model beyond the default configuration provided here.
49
 
50
+ See our repository for [EfficientNet-B4 on GitHub](https://github.com/qualcomm/ai-hub-models/blob/main/qai_hub_models/models/efficientnet_b4) for usage instructions.
51
 
52
  ## Model Details
53
 
 
63
  ## Performance Summary
64
  | Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit
65
  |---|---|---|---|---|---|---
66
+ | EfficientNet-B4 | ONNX | float | Snapdragon® X2 Elite | 1.631 ms | 45 - 45 MB | NPU
67
+ | EfficientNet-B4 | ONNX | float | Snapdragon® X Elite | 3.362 ms | 45 - 45 MB | NPU
68
+ | EfficientNet-B4 | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 2.262 ms | 0 - 130 MB | NPU
69
+ | EfficientNet-B4 | ONNX | float | Qualcomm® QCS8550 (Proxy) | 3.071 ms | 0 - 50 MB | NPU
70
+ | EfficientNet-B4 | ONNX | float | Qualcomm® QCS9075 | 4.011 ms | 0 - 4 MB | NPU
71
+ | EfficientNet-B4 | ONNX | float | Snapdragon® 8 Elite For Galaxy Mobile | 1.77 ms | 0 - 79 MB | NPU
72
+ | EfficientNet-B4 | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 1.467 ms | 0 - 76 MB | NPU
73
+ | EfficientNet-B4 | QNN_DLC | float | Snapdragon® X2 Elite | 1.965 ms | 1 - 1 MB | NPU
74
+ | EfficientNet-B4 | QNN_DLC | float | Snapdragon® X Elite | 3.65 ms | 1 - 1 MB | NPU
75
+ | EfficientNet-B4 | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 2.418 ms | 0 - 125 MB | NPU
76
+ | EfficientNet-B4 | QNN_DLC | float | Qualcomm® QCS8275 (Proxy) | 11.996 ms | 1 - 69 MB | NPU
77
+ | EfficientNet-B4 | QNN_DLC | float | Qualcomm® QCS8550 (Proxy) | 3.358 ms | 1 - 2 MB | NPU
78
+ | EfficientNet-B4 | QNN_DLC | float | Qualcomm® QCS9075 | 4.199 ms | 1 - 3 MB | NPU
79
+ | EfficientNet-B4 | QNN_DLC | float | Qualcomm® QCS8450 (Proxy) | 7.799 ms | 0 - 143 MB | NPU
80
+ | EfficientNet-B4 | QNN_DLC | float | Snapdragon® 8 Elite For Galaxy Mobile | 1.869 ms | 1 - 73 MB | NPU
81
+ | EfficientNet-B4 | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 Mobile | 1.508 ms | 1 - 73 MB | NPU
82
+ | EfficientNet-B4 | QNN_DLC | w8a16 | Snapdragon® X2 Elite | 1.656 ms | 0 - 0 MB | NPU
83
+ | EfficientNet-B4 | QNN_DLC | w8a16 | Snapdragon® X Elite | 3.837 ms | 0 - 0 MB | NPU
84
+ | EfficientNet-B4 | QNN_DLC | w8a16 | Snapdragon® 8 Gen 3 Mobile | 2.308 ms | 0 - 151 MB | NPU
85
+ | EfficientNet-B4 | QNN_DLC | w8a16 | Qualcomm® QCS6490 | 8.706 ms | 2 - 4 MB | NPU
86
+ | EfficientNet-B4 | QNN_DLC | w8a16 | Qualcomm® QCS8275 (Proxy) | 6.666 ms | 0 - 99 MB | NPU
87
+ | EfficientNet-B4 | QNN_DLC | w8a16 | Qualcomm® QCS8550 (Proxy) | 3.464 ms | 0 - 2 MB | NPU
88
+ | EfficientNet-B4 | QNN_DLC | w8a16 | Qualcomm® QCS9075 | 3.805 ms | 0 - 2 MB | NPU
89
+ | EfficientNet-B4 | QNN_DLC | w8a16 | Qualcomm® QCM6690 | 15.762 ms | 0 - 231 MB | NPU
90
+ | EfficientNet-B4 | QNN_DLC | w8a16 | Qualcomm® QCS8450 (Proxy) | 4.118 ms | 0 - 153 MB | NPU
91
+ | EfficientNet-B4 | QNN_DLC | w8a16 | Snapdragon® 8 Elite For Galaxy Mobile | 1.61 ms | 0 - 104 MB | NPU
92
+ | EfficientNet-B4 | QNN_DLC | w8a16 | Snapdragon® 7 Gen 4 Mobile | 3.597 ms | 0 - 109 MB | NPU
93
+ | EfficientNet-B4 | QNN_DLC | w8a16 | Snapdragon® 8 Elite Gen 5 Mobile | 1.321 ms | 0 - 102 MB | NPU
94
+ | EfficientNet-B4 | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 2.412 ms | 0 - 168 MB | NPU
95
+ | EfficientNet-B4 | TFLITE | float | Qualcomm® QCS8275 (Proxy) | 11.996 ms | 0 - 105 MB | NPU
96
+ | EfficientNet-B4 | TFLITE | float | Qualcomm® QCS8550 (Proxy) | 3.349 ms | 0 - 2 MB | NPU
97
+ | EfficientNet-B4 | TFLITE | float | Qualcomm® QCS9075 | 4.202 ms | 0 - 48 MB | NPU
98
+ | EfficientNet-B4 | TFLITE | float | Qualcomm® QCS8450 (Proxy) | 7.835 ms | 0 - 187 MB | NPU
99
+ | EfficientNet-B4 | TFLITE | float | Snapdragon® 8 Elite For Galaxy Mobile | 1.859 ms | 0 - 109 MB | NPU
100
+ | EfficientNet-B4 | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 1.511 ms | 0 - 107 MB | NPU
101
 
102
  ## License
103
  * The license for the original implementation of EfficientNet-B4 can be found