Hugging Face's logo

qualcomm
/
WeDetect

Object Detection
PyTorch
real_time
android
Model card Files
xet
 Community
qaihm-bot commited on
Commit
08e0b16
·
verified ·
1 Parent(s): 48a8354

v0.61.0

Browse files

See https://github.com/qualcomm/ai-hub-models/releases/v0.61.0 for changelog.

Files changed (2) hide show
  1. LICENSE +1 -0
  2. README.md +200 -0
LICENSE ADDED
@@ -0,0 +1 @@
 
 
1
+ The license of the original trained model can be found at https://github.com/WeChatCV/WeDetect/blob/main/README.md.
README.md ADDED
@@ -0,0 +1,200 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: pytorch
3
+ license: gpl-3.0
4
+ tags:
5
+ - real_time
6
+ - android
7
+ pipeline_tag: object-detection
8
+
9
+ ---
10
+
11
+ ![](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/wedetect/web-assets/model_demo.png)
12
+
13
+ # WeDetect: Optimized for Qualcomm Devices
14
+
15
+ WeDetect is a text-conditioned object detection model that uses a prompt-then-detect strategy. Text class names are reparameterized into the model weights before export, producing a fixed-vocabulary detector that runs efficiently on-device. The model uses LTRB (Left-Top-Right-Bottom) regression with multi-scale feature maps at strides 8, 16, and 32.
16
+
17
+ This is based on the implementation of WeDetect found [here](https://github.com/WeChatCV/WeDetect).
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/v0.61.0/src/qai_hub_models/models/wedetect) 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/v0.61.0/src/qai_hub_models/models/wedetect) 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 [WeDetect on GitHub](https://github.com/qualcomm/ai-hub-models/blob/v0.61.0/src/qai_hub_models/models/wedetect) for usage instructions.
30
+
31
+
32
+ ## Model Details
33
+
34
+ **Model Type:** Model_use_case.object_detection
35
+
36
+ **Model Stats:**
37
+ - Input resolution: 640x640
38
+ - Model checkpoint: wedetect_tiny
39
+ - Model size (detector) (float): 143 MB
40
+ - Model size (detector) (mixed_with_float): 35.9 MB
41
+ - Model size (text_encoder) (float): 592 B
42
+ - Model size (text_encoder) (mixed_with_float): 13.3 KB
43
+ - Number of parameters (detector): 37.3M
44
+ - Number of parameters (text_encoder): 3.00
45
+
46
+ ## Performance Summary
47
+ | Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit
48
+ |---|---|---|---|---|---|---
49
+ | detector | ONNX | float | Snapdragon® X2 Elite | 10.426 ms | 5 - 5 MB | NPU
50
+ | detector | ONNX | float | Snapdragon® X Elite | 21.594 ms | 71 - 71 MB | NPU
51
+ | detector | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 15.848 ms | 0 - 331 MB | NPU
52
+ | detector | ONNX | float | Snapdragon® 8 Gen 1 Mobile | 82.452 ms | 1 - 325 MB | NPU
53
+ | detector | ONNX | float | Qualcomm® Dragonwing™ IQ-8275 | 35.18 ms | 5 - 13 MB | NPU
54
+ | detector | ONNX | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 21.019 ms | 0 - 77 MB | NPU
55
+ | detector | ONNX | float | Qualcomm® QCS8450 | 82.452 ms | 1 - 325 MB | NPU
56
+ | detector | ONNX | float | Qualcomm® Dragonwing™ IQ-9075 | 33.506 ms | 5 - 12 MB | NPU
57
+ | detector | ONNX | float | Qualcomm® Dragonwing™ IQ-X7181 | 21.594 ms | 71 - 71 MB | NPU
58
+ | detector | ONNX | float | Qualcomm® Dragonwing™ Q-8750 | 11.352 ms | 3 - 260 MB | NPU
59
+ | detector | ONNX | float | Snapdragon® 8 Elite Mobile | 11.352 ms | 3 - 260 MB | NPU
60
+ | detector | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 8.252 ms | 1 - 262 MB | NPU
61
+ | detector | ONNX | mixed_with_float | Snapdragon® X2 Elite | 10.674 ms | 9 - 9 MB | NPU
62
+ | detector | ONNX | mixed_with_float | Snapdragon® X Elite | 16.478 ms | 35 - 35 MB | NPU
63
+ | detector | ONNX | mixed_with_float | Snapdragon® 8 Gen 3 Mobile | 11.362 ms | 2 - 365 MB | NPU
64
+ | detector | ONNX | mixed_with_float | Snapdragon® 8 Gen 1 Mobile | 25.631 ms | 2 - 363 MB | NPU
65
+ | detector | ONNX | mixed_with_float | Qualcomm® Dragonwing™ IQ-8275 | 15.652 ms | 6 - 14 MB | NPU
66
+ | detector | ONNX | mixed_with_float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 16.014 ms | 8 - 11 MB | NPU
67
+ | detector | ONNX | mixed_with_float | Qualcomm® QCS8450 | 25.631 ms | 2 - 363 MB | NPU
68
+ | detector | ONNX | mixed_with_float | Qualcomm® Dragonwing™ IQ-9075 | 17.237 ms | 6 - 14 MB | NPU
69
+ | detector | ONNX | mixed_with_float | Qualcomm® Dragonwing™ IQ-X7181 | 16.478 ms | 35 - 35 MB | NPU
70
+ | detector | ONNX | mixed_with_float | Qualcomm® Dragonwing™ Q-8750 | 11.243 ms | 8 - 314 MB | NPU
71
+ | detector | ONNX | mixed_with_float | Snapdragon® 8 Elite Mobile | 11.243 ms | 8 - 314 MB | NPU
72
+ | detector | ONNX | mixed_with_float | Snapdragon® 8 Elite Gen 5 Mobile | 9.44 ms | 4 - 316 MB | NPU
73
+ | detector | QNN_DLC | float | Snapdragon® X2 Elite | 10.354 ms | 5 - 5 MB | NPU
74
+ | detector | QNN_DLC | float | Snapdragon® X Elite | 21.052 ms | 5 - 5 MB | NPU
75
+ | detector | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 15.376 ms | 5 - 367 MB | NPU
76
+ | detector | QNN_DLC | float | Snapdragon® 8 Gen 1 Mobile | 81.716 ms | 5 - 377 MB | NPU
77
+ | detector | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-8275 | 34.453 ms | 5 - 12 MB | NPU
78
+ | detector | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-8275 | 116.247 ms | 1 - 292 MB | NPU
79
+ | detector | QNN_DLC | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 20.178 ms | 5 - 7 MB | NPU
80
+ | detector | QNN_DLC | float | Qualcomm® SA8775P | 30.364 ms | 2 - 295 MB | NPU
81
+ | detector | QNN_DLC | float | Qualcomm® SA8650P | 30.364 ms | 2 - 295 MB | NPU
82
+ | detector | QNN_DLC | float | Qualcomm® SA8255P | 30.364 ms | 2 - 295 MB | NPU
83
+ | detector | QNN_DLC | float | Qualcomm® QCS8450 | 81.716 ms | 5 - 377 MB | NPU
84
+ | detector | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-9075 | 30.784 ms | 7 - 14 MB | NPU
85
+ | detector | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-X7181 | 21.052 ms | 5 - 5 MB | NPU
86
+ | detector | QNN_DLC | float | Qualcomm® Dragonwing™ Q-8750 | 10.918 ms | 0 - 270 MB | NPU
87
+ | detector | QNN_DLC | float | Qualcomm® SA7255P | 116.247 ms | 1 - 292 MB | NPU
88
+ | detector | QNN_DLC | float | Qualcomm® SA8295P | 86.75 ms | 0 - 284 MB | NPU
89
+ | detector | QNN_DLC | float | Snapdragon® 8 Elite Mobile | 10.918 ms | 0 - 270 MB | NPU
90
+ | detector | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 Mobile | 8.078 ms | 5 - 284 MB | NPU
91
+ | detector | QNN_DLC | mixed_with_float | Snapdragon® X2 Elite | 10.056 ms | 5 - 5 MB | NPU
92
+ | detector | QNN_DLC | mixed_with_float | Snapdragon® 8 Gen 3 Mobile | 13.033 ms | 0 - 360 MB | NPU
93
+ | detector | QNN_DLC | mixed_with_float | Qualcomm® Dragonwing™ IQ-8275 | 18.18 ms | 5 - 12 MB | NPU
94
+ | detector | QNN_DLC | mixed_with_float | Qualcomm® Dragonwing™ IQ-8275 | 43.62 ms | 1 - 307 MB | NPU
95
+ | detector | QNN_DLC | mixed_with_float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 18.008 ms | 5 - 7 MB | NPU
96
+ | detector | QNN_DLC | mixed_with_float | Qualcomm® SA8775P | 18.231 ms | 0 - 307 MB | NPU
97
+ | detector | QNN_DLC | mixed_with_float | Qualcomm® SA8650P | 18.231 ms | 0 - 307 MB | NPU
98
+ | detector | QNN_DLC | mixed_with_float | Qualcomm® SA8255P | 18.231 ms | 0 - 307 MB | NPU
99
+ | detector | QNN_DLC | mixed_with_float | Qualcomm® Dragonwing™ IQ-9075 | 18.814 ms | 7 - 13 MB | NPU
100
+ | detector | QNN_DLC | mixed_with_float | Qualcomm® Dragonwing™ Q-8750 | 11.04 ms | 5 - 308 MB | NPU
101
+ | detector | QNN_DLC | mixed_with_float | Qualcomm® SA7255P | 43.62 ms | 1 - 307 MB | NPU
102
+ | detector | QNN_DLC | mixed_with_float | Snapdragon® 8 Elite Mobile | 11.04 ms | 5 - 308 MB | NPU
103
+ | detector | QNN_DLC | mixed_with_float | Snapdragon® 8 Elite Gen 5 Mobile | 8.058 ms | 5 - 326 MB | NPU
104
+ | detector | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 16.057 ms | 0 - 343 MB | NPU
105
+ | detector | TFLITE | float | Snapdragon® 8 Gen 1 Mobile | 94.737 ms | 1 - 337 MB | NPU
106
+ | detector | TFLITE | float | Qualcomm® Dragonwing™ IQ-8275 | 35.505 ms | 0 - 83 MB | NPU
107
+ | detector | TFLITE | float | Qualcomm® Dragonwing™ IQ-8275 | 117.868 ms | 1 - 266 MB | NPU
108
+ | detector | TFLITE | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 21.216 ms | 0 - 7 MB | NPU
109
+ | detector | TFLITE | float | Qualcomm® SA8775P | 31.581 ms | 0 - 268 MB | NPU
110
+ | detector | TFLITE | float | Qualcomm® SA8650P | 31.581 ms | 0 - 268 MB | NPU
111
+ | detector | TFLITE | float | Qualcomm® SA8255P | 31.581 ms | 0 - 268 MB | NPU
112
+ | detector | TFLITE | float | Qualcomm® QCS8450 | 94.737 ms | 1 - 337 MB | NPU
113
+ | detector | TFLITE | float | Qualcomm® Dragonwing™ IQ-9075 | 31.63 ms | 0 - 83 MB | NPU
114
+ | detector | TFLITE | float | Qualcomm® Dragonwing™ Q-8750 | 11.797 ms | 0 - 271 MB | NPU
115
+ | detector | TFLITE | float | Qualcomm® SA7255P | 117.868 ms | 1 - 266 MB | NPU
116
+ | detector | TFLITE | float | Qualcomm® SA8295P | 83.191 ms | 0 - 265 MB | NPU
117
+ | detector | TFLITE | float | Snapdragon® 8 Elite Mobile | 11.797 ms | 0 - 271 MB | NPU
118
+ | detector | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 9.883 ms | 0 - 273 MB | NPU
119
+ | text_encoder | ONNX | float | Snapdragon® X2 Elite | 42.773 ms | 1 - 1 MB | NPU
120
+ | text_encoder | ONNX | float | Snapdragon® X Elite | 97.059 ms | 550 - 550 MB | NPU
121
+ | text_encoder | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 65.375 ms | 1 - 2266 MB | NPU
122
+ | text_encoder | ONNX | float | Snapdragon® 8 Gen 1 Mobile | 146.223 ms | 2 - 1863 MB | NPU
123
+ | text_encoder | ONNX | float | Qualcomm® Dragonwing™ IQ-8275 | 102.452 ms | 1 - 4 MB | NPU
124
+ | text_encoder | ONNX | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 97.003 ms | 0 - 644 MB | NPU
125
+ | text_encoder | ONNX | float | Qualcomm® QCS8450 | 146.223 ms | 2 - 1863 MB | NPU
126
+ | text_encoder | ONNX | float | Qualcomm® Dragonwing™ IQ-9075 | 99.394 ms | 1 - 3 MB | NPU
127
+ | text_encoder | ONNX | float | Qualcomm® Dragonwing™ IQ-X7181 | 97.059 ms | 550 - 550 MB | NPU
128
+ | text_encoder | ONNX | float | Qualcomm® Dragonwing™ Q-8750 | 46.642 ms | 1 - 2019 MB | NPU
129
+ | text_encoder | ONNX | float | Snapdragon® 8 Elite Mobile | 46.642 ms | 1 - 2019 MB | NPU
130
+ | text_encoder | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 40.332 ms | 1 - 2264 MB | NPU
131
+ | text_encoder | ONNX | mixed_with_float | Snapdragon® X2 Elite | 42.617 ms | 1 - 1 MB | NPU
132
+ | text_encoder | ONNX | mixed_with_float | Snapdragon® X Elite | 97.181 ms | 550 - 550 MB | NPU
133
+ | text_encoder | ONNX | mixed_with_float | Snapdragon® 8 Gen 3 Mobile | 65.815 ms | 1 - 2268 MB | NPU
134
+ | text_encoder | ONNX | mixed_with_float | Snapdragon® 8 Gen 1 Mobile | 145.25 ms | 2 - 1865 MB | NPU
135
+ | text_encoder | ONNX | mixed_with_float | Qualcomm® Dragonwing™ IQ-8275 | 103.16 ms | 1 - 4 MB | NPU
136
+ | text_encoder | ONNX | mixed_with_float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 96.752 ms | 0 - 643 MB | NPU
137
+ | text_encoder | ONNX | mixed_with_float | Qualcomm® QCS8450 | 145.25 ms | 2 - 1865 MB | NPU
138
+ | text_encoder | ONNX | mixed_with_float | Qualcomm® Dragonwing™ IQ-9075 | 99.51 ms | 1 - 3 MB | NPU
139
+ | text_encoder | ONNX | mixed_with_float | Qualcomm® Dragonwing™ IQ-X7181 | 97.181 ms | 550 - 550 MB | NPU
140
+ | text_encoder | ONNX | mixed_with_float | Qualcomm® Dragonwing™ Q-8750 | 48.085 ms | 1 - 2019 MB | NPU
141
+ | text_encoder | ONNX | mixed_with_float | Snapdragon® 8 Elite Mobile | 48.085 ms | 1 - 2019 MB | NPU
142
+ | text_encoder | ONNX | mixed_with_float | Snapdragon® 8 Elite Gen 5 Mobile | 39.855 ms | 1 - 2265 MB | NPU
143
+ | text_encoder | QNN_DLC | float | Snapdragon® X2 Elite | 43.277 ms | 0 - 0 MB | NPU
144
+ | text_encoder | QNN_DLC | float | Snapdragon® X Elite | 86.461 ms | 0 - 0 MB | NPU
145
+ | text_encoder | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 62.016 ms | 0 - 2450 MB | NPU
146
+ | text_encoder | QNN_DLC | float | Snapdragon® 8 Gen 1 Mobile | 140.009 ms | 0 - 3013 MB | NPU
147
+ | text_encoder | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-8275 | 95.48 ms | 0 - 3 MB | NPU
148
+ | text_encoder | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-8275 | 260.992 ms | 1 - 2036 MB | NPU
149
+ | text_encoder | QNN_DLC | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 85.173 ms | 0 - 3 MB | NPU
150
+ | text_encoder | QNN_DLC | float | Qualcomm® SA8775P | 95.567 ms | 1 - 2019 MB | NPU
151
+ | text_encoder | QNN_DLC | float | Qualcomm® SA8650P | 95.567 ms | 1 - 2019 MB | NPU
152
+ | text_encoder | QNN_DLC | float | Qualcomm® SA8255P | 95.567 ms | 1 - 2019 MB | NPU
153
+ | text_encoder | QNN_DLC | float | Qualcomm® QCS8450 | 140.009 ms | 0 - 3013 MB | NPU
154
+ | text_encoder | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-9075 | 93.955 ms | 0 - 2 MB | NPU
155
+ | text_encoder | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-X7181 | 86.461 ms | 0 - 0 MB | NPU
156
+ | text_encoder | QNN_DLC | float | Qualcomm® Dragonwing™ Q-8750 | 46.578 ms | 0 - 1973 MB | NPU
157
+ | text_encoder | QNN_DLC | float | Qualcomm® SA7255P | 260.992 ms | 1 - 2036 MB | NPU
158
+ | text_encoder | QNN_DLC | float | Qualcomm® SA8295P | 107.476 ms | 0 - 1902 MB | NPU
159
+ | text_encoder | QNN_DLC | float | Snapdragon® 8 Elite Mobile | 46.578 ms | 0 - 1973 MB | NPU
160
+ | text_encoder | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 Mobile | 34.527 ms | 0 - 1983 MB | NPU
161
+ | text_encoder | QNN_DLC | mixed_with_float | Snapdragon® X2 Elite | 37.173 ms | 0 - 0 MB | NPU
162
+ | text_encoder | QNN_DLC | mixed_with_float | Snapdragon® 8 Gen 3 Mobile | 61.867 ms | 0 - 2451 MB | NPU
163
+ | text_encoder | QNN_DLC | mixed_with_float | Qualcomm® Dragonwing™ IQ-8275 | 95.226 ms | 0 - 3 MB | NPU
164
+ | text_encoder | QNN_DLC | mixed_with_float | Qualcomm® Dragonwing™ IQ-8275 | 261.037 ms | 0 - 2035 MB | NPU
165
+ | text_encoder | QNN_DLC | mixed_with_float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 85.295 ms | 0 - 4 MB | NPU
166
+ | text_encoder | QNN_DLC | mixed_with_float | Qualcomm® SA8775P | 95.609 ms | 1 - 2019 MB | NPU
167
+ | text_encoder | QNN_DLC | mixed_with_float | Qualcomm® SA8650P | 95.609 ms | 1 - 2019 MB | NPU
168
+ | text_encoder | QNN_DLC | mixed_with_float | Qualcomm® SA8255P | 95.609 ms | 1 - 2019 MB | NPU
169
+ | text_encoder | QNN_DLC | mixed_with_float | Qualcomm® Dragonwing™ IQ-9075 | 93.359 ms | 0 - 2 MB | NPU
170
+ | text_encoder | QNN_DLC | mixed_with_float | Qualcomm® Dragonwing™ Q-8750 | 47.44 ms | 1 - 1975 MB | NPU
171
+ | text_encoder | QNN_DLC | mixed_with_float | Qualcomm® SA7255P | 261.037 ms | 0 - 2035 MB | NPU
172
+ | text_encoder | QNN_DLC | mixed_with_float | Snapdragon® 8 Elite Mobile | 47.44 ms | 1 - 1975 MB | NPU
173
+ | text_encoder | QNN_DLC | mixed_with_float | Snapdragon® 8 Elite Gen 5 Mobile | 34.504 ms | 0 - 1983 MB | NPU
174
+ | text_encoder | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 72.829 ms | 1 - 3683 MB | NPU
175
+ | text_encoder | TFLITE | float | Snapdragon® 8 Gen 1 Mobile | 159.932 ms | 1 - 3564 MB | NPU
176
+ | text_encoder | TFLITE | float | Qualcomm® Dragonwing™ IQ-8275 | 109.605 ms | 0 - 553 MB | NPU
177
+ | text_encoder | TFLITE | float | Qualcomm® Dragonwing™ IQ-8275 | 289.737 ms | 1 - 2144 MB | NPU
178
+ | text_encoder | TFLITE | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 99.966 ms | 1 - 4 MB | NPU
179
+ | text_encoder | TFLITE | float | Qualcomm® SA8775P | 111.295 ms | 1 - 2144 MB | NPU
180
+ | text_encoder | TFLITE | float | Qualcomm® SA8650P | 111.295 ms | 1 - 2144 MB | NPU
181
+ | text_encoder | TFLITE | float | Qualcomm® SA8255P | 111.295 ms | 1 - 2144 MB | NPU
182
+ | text_encoder | TFLITE | float | Qualcomm® QCS8450 | 159.932 ms | 1 - 3564 MB | NPU
183
+ | text_encoder | TFLITE | float | Qualcomm® Dragonwing™ IQ-9075 | 108.948 ms | 0 - 553 MB | NPU
184
+ | text_encoder | TFLITE | float | Qualcomm® Dragonwing™ Q-8750 | 55.49 ms | 1 - 2073 MB | NPU
185
+ | text_encoder | TFLITE | float | Qualcomm® SA7255P | 289.737 ms | 1 - 2144 MB | NPU
186
+ | text_encoder | TFLITE | float | Qualcomm® SA8295P | 112.85 ms | 1 - 2099 MB | NPU
187
+ | text_encoder | TFLITE | float | Snapdragon® 8 Elite Mobile | 55.49 ms | 1 - 2073 MB | NPU
188
+ | text_encoder | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 41.511 ms | 0 - 2220 MB | NPU
189
+
190
+ ## License
191
+ * The license for the original implementation of WeDetect can be found
192
+ [here](https://github.com/WeChatCV/WeDetect/blob/main/README.md).
193
+
194
+ ## References
195
+ * [WeDetect: Fast Open-Vocabulary Object Detection as Retrieval](https://arxiv.org/abs/2512.12309)
196
+ * [Source Model Implementation](https://github.com/WeChatCV/WeDetect)
197
+
198
+ ## Community
199
+ * Join [our AI Hub Slack community](https://aihub.qualcomm.com/community/slack) to collaborate, post questions and learn more about on-device AI.
200
+ * For questions or feedback please [reach out to us](mailto:ai-hub-support@qti.qualcomm.com).