Hugging Face's logo

qualcomm
/
Yolo-v7

Object Detection
PyTorch
TF Lite
ONNX
real_time
android
Model card Files Community
qaihm-bot commited on
Commit
3dceede
·
verified ·
1 Parent(s): 798bedf

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +40 -19
README.md CHANGED
@@ -15,7 +15,7 @@ tags:
15
 
16
  YoloV7 is a machine learning model that predicts bounding boxes and classes of objects in an image.
17
 
18
- This model is an implementation of Yolo-v7 found [here](https://github.com/WongKinYiu/yolov7/).
19
  This repository provides scripts to run Yolo-v7 on Qualcomm® devices.
20
  More details on model performance across various devices, can be found
21
  [here](https://aihub.qualcomm.com/models/yolov7).
@@ -30,15 +30,32 @@ More details on model performance across various devices, can be found
30
  - Number of parameters: 6.39M
31
  - Model size: 24.4 MB
32
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
33
 
34
 
35
 
36
- | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Precision | Primary Compute Unit | Target Model
37
- | ---|---|---|---|---|---|---|---|
38
- | Samsung Galaxy S23 Ultra (Android 13) | Snapdragon® 8 Gen 2 | TFLite | 17.219 ms | 0 - 3 MB | FP16 | NPU | [Yolo-v7.tflite](https://huggingface.co/qualcomm/Yolo-v7/blob/main/Yolo-v7.tflite)
39
- | Samsung Galaxy S23 Ultra (Android 13) | Snapdragon® 8 Gen 2 | QNN Model Library | 10.503 ms | 5 - 18 MB | FP16 | NPU | [Yolo-v7.so](https://huggingface.co/qualcomm/Yolo-v7/blob/main/Yolo-v7.so)
40
-
41
-
42
 
43
  ## Installation
44
 
@@ -94,16 +111,16 @@ device. This script does the following:
94
  ```bash
95
  python -m qai_hub_models.models.yolov7.export
96
  ```
97
-
98
  ```
99
- Profile Job summary of Yolo-v7
100
- --------------------------------------------------
101
- Device: Snapdragon X Elite CRD (11)
102
- Estimated Inference Time: 10.92 ms
103
- Estimated Peak Memory Range: 4.70-4.70 MB
104
- Compute Units: NPU (221) | Total (221)
105
-
106
-
 
107
  ```
108
 
109
 
@@ -202,15 +219,19 @@ provides instructions on how to use the `.so` shared library in an Android appl
202
  Get more details on Yolo-v7's performance across various devices [here](https://aihub.qualcomm.com/models/yolov7).
203
  Explore all available models on [Qualcomm® AI Hub](https://aihub.qualcomm.com/)
204
 
 
205
  ## License
206
- - The license for the original implementation of Yolo-v7 can be found
207
- [here](https://github.com/WongKinYiu/yolov7/blob/main/LICENSE.md).
208
- - The license for the compiled assets for on-device deployment can be found [here](https://github.com/WongKinYiu/yolov7/blob/main/LICENSE.md)
 
209
 
210
  ## References
211
  * [YOLOv7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors](https://arxiv.org/abs/2207.02696)
212
  * [Source Model Implementation](https://github.com/WongKinYiu/yolov7/)
213
 
 
 
214
  ## Community
215
  * Join [our AI Hub Slack community](https://aihub.qualcomm.com/community/slack) to collaborate, post questions and learn more about on-device AI.
216
  * For questions or feedback please [reach out to us](mailto:[email protected]).
 
15
 
16
  YoloV7 is a machine learning model that predicts bounding boxes and classes of objects in an image.
17
 
18
+ This model is an implementation of Yolo-v7 found [here]({source_repo}).
19
  This repository provides scripts to run Yolo-v7 on Qualcomm® devices.
20
  More details on model performance across various devices, can be found
21
  [here](https://aihub.qualcomm.com/models/yolov7).
 
30
  - Number of parameters: 6.39M
31
  - Model size: 24.4 MB
32
 
33
+ | Model | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Precision | Primary Compute Unit | Target Model
34
+ |---|---|---|---|---|---|---|---|---|
35
+ | Yolo-v7 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | TFLITE | 17.188 ms | 1 - 3 MB | FP16 | NPU | [Yolo-v7.tflite](https://huggingface.co/qualcomm/Yolo-v7/blob/main/Yolo-v7.tflite) |
36
+ | Yolo-v7 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | QNN | 10.527 ms | 5 - 21 MB | FP16 | NPU | [Yolo-v7.so](https://huggingface.co/qualcomm/Yolo-v7/blob/main/Yolo-v7.so) |
37
+ | Yolo-v7 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | ONNX | 12.235 ms | 0 - 12 MB | FP16 | NPU | [Yolo-v7.onnx](https://huggingface.co/qualcomm/Yolo-v7/blob/main/Yolo-v7.onnx) |
38
+ | Yolo-v7 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | TFLITE | 11.658 ms | 1 - 101 MB | FP16 | NPU | [Yolo-v7.tflite](https://huggingface.co/qualcomm/Yolo-v7/blob/main/Yolo-v7.tflite) |
39
+ | Yolo-v7 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | QNN | 7.221 ms | 5 - 75 MB | FP16 | NPU | [Yolo-v7.so](https://huggingface.co/qualcomm/Yolo-v7/blob/main/Yolo-v7.so) |
40
+ | Yolo-v7 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | ONNX | 8.179 ms | 0 - 119 MB | FP16 | NPU | [Yolo-v7.onnx](https://huggingface.co/qualcomm/Yolo-v7/blob/main/Yolo-v7.onnx) |
41
+ | Yolo-v7 | QCS8550 (Proxy) | QCS8550 Proxy | TFLITE | 17.172 ms | 1 - 9 MB | FP16 | NPU | [Yolo-v7.tflite](https://huggingface.co/qualcomm/Yolo-v7/blob/main/Yolo-v7.tflite) |
42
+ | Yolo-v7 | QCS8550 (Proxy) | QCS8550 Proxy | QNN | 10.322 ms | 5 - 6 MB | FP16 | NPU | Use Export Script |
43
+ | Yolo-v7 | SA8255 (Proxy) | SA8255P Proxy | TFLITE | 17.145 ms | 0 - 2 MB | FP16 | NPU | [Yolo-v7.tflite](https://huggingface.co/qualcomm/Yolo-v7/blob/main/Yolo-v7.tflite) |
44
+ | Yolo-v7 | SA8255 (Proxy) | SA8255P Proxy | QNN | 10.334 ms | 5 - 6 MB | FP16 | NPU | Use Export Script |
45
+ | Yolo-v7 | SA8775 (Proxy) | SA8775P Proxy | TFLITE | 17.142 ms | 0 - 2 MB | FP16 | NPU | [Yolo-v7.tflite](https://huggingface.co/qualcomm/Yolo-v7/blob/main/Yolo-v7.tflite) |
46
+ | Yolo-v7 | SA8775 (Proxy) | SA8775P Proxy | QNN | 10.477 ms | 5 - 6 MB | FP16 | NPU | Use Export Script |
47
+ | Yolo-v7 | SA8650 (Proxy) | SA8650P Proxy | TFLITE | 17.156 ms | 0 - 202 MB | FP16 | NPU | [Yolo-v7.tflite](https://huggingface.co/qualcomm/Yolo-v7/blob/main/Yolo-v7.tflite) |
48
+ | Yolo-v7 | SA8650 (Proxy) | SA8650P Proxy | QNN | 10.469 ms | 5 - 6 MB | FP16 | NPU | Use Export Script |
49
+ | Yolo-v7 | QCS8450 (Proxy) | QCS8450 Proxy | TFLITE | 19.533 ms | 1 - 102 MB | FP16 | NPU | [Yolo-v7.tflite](https://huggingface.co/qualcomm/Yolo-v7/blob/main/Yolo-v7.tflite) |
50
+ | Yolo-v7 | QCS8450 (Proxy) | QCS8450 Proxy | QNN | 12.605 ms | 5 - 61 MB | FP16 | NPU | Use Export Script |
51
+ | Yolo-v7 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | TFLITE | 12.241 ms | 1 - 69 MB | FP16 | NPU | [Yolo-v7.tflite](https://huggingface.co/qualcomm/Yolo-v7/blob/main/Yolo-v7.tflite) |
52
+ | Yolo-v7 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | QNN | 5.825 ms | 5 - 70 MB | FP16 | NPU | Use Export Script |
53
+ | Yolo-v7 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | ONNX | 8.115 ms | 6 - 86 MB | FP16 | NPU | [Yolo-v7.onnx](https://huggingface.co/qualcomm/Yolo-v7/blob/main/Yolo-v7.onnx) |
54
+ | Yolo-v7 | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN | 10.949 ms | 5 - 5 MB | FP16 | NPU | Use Export Script |
55
+ | Yolo-v7 | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 14.157 ms | 9 - 9 MB | FP16 | NPU | [Yolo-v7.onnx](https://huggingface.co/qualcomm/Yolo-v7/blob/main/Yolo-v7.onnx) |
56
 
57
 
58
 
 
 
 
 
 
 
59
 
60
  ## Installation
61
 
 
111
  ```bash
112
  python -m qai_hub_models.models.yolov7.export
113
  ```
 
114
  ```
115
+ Profiling Results
116
+ ------------------------------------------------------------
117
+ Yolo-v7
118
+ Device : Samsung Galaxy S23 (13)
119
+ Runtime : TFLITE
120
+ Estimated inference time (ms) : 17.2
121
+ Estimated peak memory usage (MB): [1, 3]
122
+ Total # Ops : 215
123
+ Compute Unit(s) : NPU (215 ops)
124
  ```
125
 
126
 
 
219
  Get more details on Yolo-v7's performance across various devices [here](https://aihub.qualcomm.com/models/yolov7).
220
  Explore all available models on [Qualcomm® AI Hub](https://aihub.qualcomm.com/)
221
 
222
+
223
  ## License
224
+ * The license for the original implementation of Yolo-v7 can be found [here](https://github.com/WongKinYiu/yolov7/blob/main/LICENSE.md).
225
+ * The license for the compiled assets for on-device deployment can be found [here](https://github.com/WongKinYiu/yolov7/blob/main/LICENSE.md)
226
+
227
+
228
 
229
  ## References
230
  * [YOLOv7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors](https://arxiv.org/abs/2207.02696)
231
  * [Source Model Implementation](https://github.com/WongKinYiu/yolov7/)
232
 
233
+
234
+
235
  ## Community
236
  * Join [our AI Hub Slack community](https://aihub.qualcomm.com/community/slack) to collaborate, post questions and learn more about on-device AI.
237
  * For questions or feedback please [reach out to us](mailto:[email protected]).