Hugging Face's logo

qualcomm
/
RTMDet

Object Detection
PyTorch
real_time
android
Model card Files
xet
 Community
qaihm-bot commited on
Commit
7ebe21d
·
verified ·
1 Parent(s): c75f641

v0.60.0

Browse files

See https://github.com/qualcomm/ai-hub-models/releases/v0.60.0 for changelog.

Files changed (1) hide show
  1. README.md +48 -42
README.md CHANGED
@@ -15,18 +15,18 @@ pipeline_tag: object-detection
15
  RTMDet is a highly efficient model for real-time object detection,capable of predicting both the bounding boxes and classes of objects within an image.It is highly optimized for real-time applications, making it reliable for industrial and commercial use
16
 
17
  This is based on the implementation of RTMDet found [here](https://github.com/open-mmlab/mmdetection/tree/3.x/configs/rtmdet).
18
- 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/v0.59.0/src/qai_hub_models/models/rtmdet) library to export with custom configurations. More details on model performance across various devices, can be found [here](#performance-summary).
19
 
20
  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.
21
 
22
  ## Getting Started
23
  Due to licensing restrictions, we cannot distribute pre-exported model assets for this model.
24
- Use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/v0.59.0/src/qai_hub_models/models/rtmdet) Python library to compile and export the model with your own:
25
  - Custom weights (e.g., fine-tuned checkpoints)
26
  - Custom input shapes
27
  - Target device and runtime configurations
28
 
29
- See our repository for [RTMDet on GitHub](https://github.com/qualcomm/ai-hub-models/blob/v0.59.0/src/qai_hub_models/models/rtmdet) for usage instructions.
30
 
31
 
32
  ## Model Details
@@ -34,51 +34,57 @@ See our repository for [RTMDet on GitHub](https://github.com/qualcomm/ai-hub-mod
34
  **Model Type:** Model_use_case.object_detection
35
 
36
  **Model Stats:**
37
- - Model checkpoint: RTMDet Medium
38
  - Input resolution: 640x640
39
- - Number of parameters: 27.5M
40
  - Model size (float): 105 MB
 
41
 
42
  ## Performance Summary
43
  | Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit
44
  |---|---|---|---|---|---|---
45
- | RTMDet | ONNX | float | Snapdragon® X2 Elite | 8.097 ms | 5 - 5 MB | NPU
46
- | RTMDet | ONNX | float | Snapdragon® X Elite | 15.379 ms | 51 - 51 MB | NPU
47
- | RTMDet | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 11.157 ms | 3 - 233 MB | NPU
48
- | RTMDet | ONNX | float | Snapdragon® 8 Gen 1 Mobile | 45.676 ms | 5 - 299 MB | NPU
49
- | RTMDet | ONNX | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 14.707 ms | 5 - 224 MB | NPU
50
- | RTMDet | ONNX | float | Qualcomm® QCS8450 | 45.676 ms | 5 - 299 MB | NPU
51
- | RTMDet | ONNX | float | Qualcomm® Dragonwing™ IQ-9075 | 22.885 ms | 5 - 12 MB | NPU
52
- | RTMDet | ONNX | float | Qualcomm® Dragonwing™ IQ-X7181 | 15.379 ms | 51 - 51 MB | NPU
53
- | RTMDet | ONNX | float | Qualcomm® Dragonwing™ Q-8750 | 9.29 ms | 3 - 187 MB | NPU
54
- | RTMDet | ONNX | float | Snapdragon® 8 Elite Mobile | 9.29 ms | 3 - 187 MB | NPU
55
- | RTMDet | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 6.033 ms | 2 - 192 MB | NPU
56
- | RTMDet | ONNX | w8a16 | Snapdragon® X2 Elite | 5.757 ms | 2 - 2 MB | NPU
57
- | RTMDet | ONNX | w8a16 | Snapdragon® X Elite | 14.194 ms | 27 - 27 MB | NPU
58
- | RTMDet | ONNX | w8a16 | Snapdragon® 8 Gen 3 Mobile | 8.857 ms | 3 - 356 MB | NPU
59
- | RTMDet | ONNX | w8a16 | Snapdragon® 8 Gen 1 Mobile | 18.681 ms | 3 - 358 MB | NPU
60
- | RTMDet | ONNX | w8a16 | Qualcomm® Dragonwing™ QCS6490 | 64.074 ms | 2 - 5 MB | NPU
61
- | RTMDet | ONNX | w8a16 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 13.531 ms | 2 - 5 MB | NPU
62
- | RTMDet | ONNX | w8a16 | Qualcomm® QCS8450 | 18.681 ms | 3 - 358 MB | NPU
63
- | RTMDet | ONNX | w8a16 | Qualcomm® Dragonwing™ IQ-9075 | 14.281 ms | 2 - 5 MB | NPU
64
- | RTMDet | ONNX | w8a16 | Qualcomm® Dragonwing™ IQ-X7181 | 14.194 ms | 27 - 27 MB | NPU
65
- | RTMDet | ONNX | w8a16 | Qualcomm® Dragonwing™ Q-8750 | 6.175 ms | 1 - 297 MB | NPU
66
- | RTMDet | ONNX | w8a16 | Snapdragon® 8 Elite Mobile | 6.175 ms | 1 - 297 MB | NPU
67
- | RTMDet | ONNX | w8a16 | Snapdragon® 8 Elite Gen 5 Mobile | 5.111 ms | 1 - 324 MB | NPU
68
- | RTMDet | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 20.007 ms | 0 - 305 MB | NPU
69
- | RTMDet | TFLITE | float | Snapdragon® 8 Gen 1 Mobile | 51.172 ms | 2 - 367 MB | NPU
70
- | RTMDet | TFLITE | float | Qualcomm® Dragonwing™ QCS8275 | 105.943 ms | 1 - 224 MB | NPU
71
- | RTMDet | TFLITE | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 28.927 ms | 0 - 3 MB | NPU
72
- | RTMDet | TFLITE | float | Qualcomm® SA8775P | 36.501 ms | 0 - 225 MB | NPU
73
- | RTMDet | TFLITE | float | Qualcomm® SA8650P | 36.501 ms | 0 - 225 MB | NPU
74
- | RTMDet | TFLITE | float | Qualcomm® SA8255P | 36.501 ms | 0 - 225 MB | NPU
75
- | RTMDet | TFLITE | float | Qualcomm® QCS8450 | 51.172 ms | 2 - 367 MB | NPU
76
- | RTMDet | TFLITE | float | Qualcomm® Dragonwing™ IQ-9075 | 36.373 ms | 0 - 62 MB | NPU
77
- | RTMDet | TFLITE | float | Qualcomm® Dragonwing™ Q-8750 | 14.951 ms | 0 - 228 MB | NPU
78
- | RTMDet | TFLITE | float | Qualcomm® SA7255P | 105.943 ms | 1 - 224 MB | NPU
79
- | RTMDet | TFLITE | float | Qualcomm® SA8295P | 46.064 ms | 0 - 283 MB | NPU
80
- | RTMDet | TFLITE | float | Snapdragon® 8 Elite Mobile | 14.951 ms | 0 - 228 MB | NPU
81
- | RTMDet | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 9.889 ms | 0 - 229 MB | NPU
 
 
 
 
 
 
82
 
83
  ## License
84
  * The license for the original implementation of RTMDet can be found
 
15
  RTMDet is a highly efficient model for real-time object detection,capable of predicting both the bounding boxes and classes of objects within an image.It is highly optimized for real-time applications, making it reliable for industrial and commercial use
16
 
17
  This is based on the implementation of RTMDet found [here](https://github.com/open-mmlab/mmdetection/tree/3.x/configs/rtmdet).
18
+ 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/v0.60.0/src/qai_hub_models/models/rtmdet) library to export with custom configurations. More details on model performance across various devices, can be found [here](#performance-summary).
19
 
20
  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.
21
 
22
  ## Getting Started
23
  Due to licensing restrictions, we cannot distribute pre-exported model assets for this model.
24
+ Use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/v0.60.0/src/qai_hub_models/models/rtmdet) Python library to compile and export the model with your own:
25
  - Custom weights (e.g., fine-tuned checkpoints)
26
  - Custom input shapes
27
  - Target device and runtime configurations
28
 
29
+ See our repository for [RTMDet on GitHub](https://github.com/qualcomm/ai-hub-models/blob/v0.60.0/src/qai_hub_models/models/rtmdet) for usage instructions.
30
 
31
 
32
  ## Model Details
 
34
  **Model Type:** Model_use_case.object_detection
35
 
36
  **Model Stats:**
 
37
  - Input resolution: 640x640
38
+ - Model checkpoint: RTMDet Medium
39
  - Model size (float): 105 MB
40
+ - Number of parameters: 27.5M
41
 
42
  ## Performance Summary
43
  | Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit
44
  |---|---|---|---|---|---|---
45
+ | RTMDet | ONNX | float | Snapdragon® X2 Elite | 9.048 ms | 5 - 5 MB | NPU
46
+ | RTMDet | ONNX | float | Snapdragon® X Elite | 16.602 ms | 51 - 51 MB | NPU
47
+ | RTMDet | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 12.507 ms | 5 - 242 MB | NPU
48
+ | RTMDet | ONNX | float | Snapdragon® 8 Gen 1 Mobile | 39.877 ms | 5 - 304 MB | NPU
49
+ | RTMDet | ONNX | float | Qualcomm® Dragonwing™ IQ-8275 | 27.333 ms | 5 - 13 MB | NPU
50
+ | RTMDet | ONNX | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 16.316 ms | 5 - 233 MB | NPU
51
+ | RTMDet | ONNX | float | Qualcomm® QCS8450 | 39.877 ms | 5 - 304 MB | NPU
52
+ | RTMDet | ONNX | float | Qualcomm® Dragonwing™ IQ-9075 | 24.992 ms | 5 - 12 MB | NPU
53
+ | RTMDet | ONNX | float | Qualcomm® Dragonwing™ IQ-X7181 | 16.602 ms | 51 - 51 MB | NPU
54
+ | RTMDet | ONNX | float | Qualcomm® Dragonwing™ Q-8750 | 9.53 ms | 3 - 195 MB | NPU
55
+ | RTMDet | ONNX | float | Snapdragon® 8 Elite Mobile | 9.53 ms | 3 - 195 MB | NPU
56
+ | RTMDet | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 7.1 ms | 3 - 200 MB | NPU
57
+ | RTMDet | ONNX | w8a16 | Snapdragon® X2 Elite | 6.677 ms | 2 - 2 MB | NPU
58
+ | RTMDet | ONNX | w8a16 | Snapdragon® X Elite | 15.879 ms | 27 - 27 MB | NPU
59
+ | RTMDet | ONNX | w8a16 | Snapdragon® 8 Gen 3 Mobile | 9.943 ms | 3 - 356 MB | NPU
60
+ | RTMDet | ONNX | w8a16 | Snapdragon® 8 Gen 1 Mobile | 20.486 ms | 3 - 358 MB | NPU
61
+ | RTMDet | ONNX | w8a16 | Qualcomm® Dragonwing™ QCS6490 | 65.678 ms | 26 - 28 MB | NPU
62
+ | RTMDet | ONNX | w8a16 | Qualcomm® Dragonwing™ IQ-8275 | 14.19 ms | 2 - 6 MB | NPU
63
+ | RTMDet | ONNX | w8a16 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 15.155 ms | 2 - 6 MB | NPU
64
+ | RTMDet | ONNX | w8a16 | Qualcomm® QCS8450 | 20.486 ms | 3 - 358 MB | NPU
65
+ | RTMDet | ONNX | w8a16 | Qualcomm® Dragonwing™ IQ-9075 | 15.631 ms | 2 - 5 MB | NPU
66
+ | RTMDet | ONNX | w8a16 | Qualcomm® Dragonwing™ IQ-X7181 | 15.879 ms | 27 - 27 MB | NPU
67
+ | RTMDet | ONNX | w8a16 | Qualcomm® Dragonwing™ Q-6690 | 142.898 ms | 26 - 347 MB | NPU
68
+ | RTMDet | ONNX | w8a16 | Qualcomm® Dragonwing™ Q-7790 | 22.749 ms | 25 - 352 MB | NPU
69
+ | RTMDet | ONNX | w8a16 | Qualcomm® Dragonwing™ Q-8750 | 7.229 ms | 1 - 300 MB | NPU
70
+ | RTMDet | ONNX | w8a16 | Snapdragon® 8 Elite Mobile | 7.229 ms | 1 - 300 MB | NPU
71
+ | RTMDet | ONNX | w8a16 | Snapdragon® 8 Elite Gen 5 Mobile | 5.857 ms | 0 - 328 MB | NPU
72
+ | RTMDet | ONNX | w8a16 | Snapdragon® 7 Gen 4 Mobile | 22.749 ms | 25 - 352 MB | NPU
73
+ | RTMDet | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 12.129 ms | 5 - 294 MB | NPU
74
+ | RTMDet | TFLITE | float | Snapdragon® 8 Gen 1 Mobile | 46.529 ms | 10 - 362 MB | NPU
75
+ | RTMDet | TFLITE | float | Qualcomm® Dragonwing™ IQ-8275 | 27.335 ms | 9 - 81 MB | NPU
76
+ | RTMDet | TFLITE | float | Qualcomm® Dragonwing™ IQ-8275 | 85.465 ms | 10 - 221 MB | NPU
77
+ | RTMDet | TFLITE | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 15.911 ms | 9 - 12 MB | NPU
78
+ | RTMDet | TFLITE | float | Qualcomm® SA8775P | 24.25 ms | 9 - 222 MB | NPU
79
+ | RTMDet | TFLITE | float | Qualcomm® SA8650P | 24.25 ms | 9 - 222 MB | NPU
80
+ | RTMDet | TFLITE | float | Qualcomm® SA8255P | 24.25 ms | 9 - 222 MB | NPU
81
+ | RTMDet | TFLITE | float | Qualcomm® QCS8450 | 46.529 ms | 10 - 362 MB | NPU
82
+ | RTMDet | TFLITE | float | Qualcomm® Dragonwing™ IQ-9075 | 24.34 ms | 8 - 79 MB | NPU
83
+ | RTMDet | TFLITE | float | Qualcomm® Dragonwing™ Q-8750 | 9.088 ms | 11 - 228 MB | NPU
84
+ | RTMDet | TFLITE | float | Qualcomm® SA7255P | 85.465 ms | 10 - 221 MB | NPU
85
+ | RTMDet | TFLITE | float | Qualcomm® SA8295P | 33.04 ms | 9 - 283 MB | NPU
86
+ | RTMDet | TFLITE | float | Snapdragon® 8 Elite Mobile | 9.088 ms | 11 - 228 MB | NPU
87
+ | RTMDet | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 6.903 ms | 8 - 224 MB | NPU
88
 
89
  ## License
90
  * The license for the original implementation of RTMDet can be found