Hugging Face's logo

qualcomm
/
Segment-Anything-Model

Image Segmentation
PyTorch
TF Lite
ONNX
foundation
android
Model card Files Community
qaihm-bot commited on
Commit
a04831d
·
verified ·
1 Parent(s): 7521d92

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +365 -59
README.md CHANGED
@@ -35,24 +35,150 @@ More details on model performance across various devices, can be found
35
 
36
  | Model | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Precision | Primary Compute Unit | Target Model
37
  |---|---|---|---|---|---|---|---|---|
38
- | SAMDecoder | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | TFLITE | 29.311 ms | 4 - 40 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMDecoder.tflite) |
39
- | SAMDecoder | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | TFLITE | 20.843 ms | 4 - 119 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMDecoder.tflite) |
40
- | SAMDecoder | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | TFLITE | 20.649 ms | 3 - 128 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMDecoder.tflite) |
41
- | SAMDecoder | QCS8550 (Proxy) | QCS8550 Proxy | TFLITE | 29.453 ms | 4 - 41 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMDecoder.tflite) |
42
- | SAMDecoder | SA8255 (Proxy) | SA8255P Proxy | TFLITE | 29.485 ms | 4 - 36 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMDecoder.tflite) |
43
- | SAMDecoder | SA8295P ADP | SA8295P | TFLITE | 36.583 ms | 4 - 114 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMDecoder.tflite) |
44
- | SAMDecoder | SA8650 (Proxy) | SA8650P Proxy | TFLITE | 29.064 ms | 4 - 37 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMDecoder.tflite) |
45
- | SAMDecoder | QCS8450 (Proxy) | QCS8450 Proxy | TFLITE | 33.135 ms | 4 - 126 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMDecoder.tflite) |
46
- | SAMEncoder | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | TFLITE | 10369.27 ms | 123 - 126 MB | FP32 | CPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoder.tflite) |
47
- | SAMEncoder | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | TFLITE | 9136.859 ms | 91 - 1603 MB | FP32 | CPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoder.tflite) |
48
- | SAMEncoder | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | TFLITE | 6694.41 ms | 110 - 1589 MB | FP32 | CPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoder.tflite) |
49
- | SAMEncoder | QCS8550 (Proxy) | QCS8550 Proxy | TFLITE | 10714.354 ms | 122 - 381 MB | FP32 | CPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoder.tflite) |
50
- | SAMEncoder | SA7255P ADP | SA7255P | TFLITE | 17192.024 ms | 125 - 1597 MB | FP32 | CPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoder.tflite) |
51
- | SAMEncoder | SA8255 (Proxy) | SA8255P Proxy | TFLITE | 11081.197 ms | 124 - 127 MB | FP32 | CPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoder.tflite) |
52
- | SAMEncoder | SA8295P ADP | SA8295P | TFLITE | 10359.771 ms | 124 - 1654 MB | FP32 | CPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoder.tflite) |
53
- | SAMEncoder | SA8650 (Proxy) | SA8650P Proxy | TFLITE | 10872.614 ms | 118 - 125 MB | FP32 | CPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoder.tflite) |
54
- | SAMEncoder | SA8775P ADP | SA8775P | TFLITE | 13361.634 ms | 124 - 1596 MB | FP32 | CPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoder.tflite) |
55
- | SAMEncoder | QCS8450 (Proxy) | QCS8450 Proxy | TFLITE | 16830.94 ms | 82 - 1648 MB | FP32 | CPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoder.tflite) |
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
56
 
57
 
58
 
@@ -117,19 +243,64 @@ Profiling Results
117
  SAMDecoder
118
  Device : Samsung Galaxy S23 (13)
119
  Runtime : TFLITE
120
- Estimated inference time (ms) : 29.3
121
- Estimated peak memory usage (MB): [4, 40]
122
- Total # Ops : 337
123
- Compute Unit(s) : NPU (337 ops)
124
 
125
  ------------------------------------------------------------
126
- SAMEncoder
127
- Device : Samsung Galaxy S23 (13)
128
- Runtime : TFLITE
129
- Estimated inference time (ms) : 10369.3
130
- Estimated peak memory usage (MB): [123, 126]
131
- Total # Ops : 818
132
- Compute Unit(s) : GPU (36 ops) CPU (782 ops)
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
133
  ```
134
 
135
 
@@ -152,42 +323,122 @@ from qai_hub_models.models.sam import Model
152
 
153
  # Load the model
154
  model = Model.from_pretrained()
155
- get_sam_decoder()_model = model.get_sam_decoder()
156
- get_sam_encoder()_model = model.get_sam_encoder()
 
 
 
 
 
157
 
158
  # Device
159
  device = hub.Device("Samsung Galaxy S23")
160
 
161
  # Trace model
162
- get_sam_decoder()_input_shape = get_sam_decoder()_model.get_input_spec()
163
- get_sam_decoder()_sample_inputs = get_sam_decoder()_model.sample_inputs()
164
 
165
- traced_get_sam_decoder()_model = torch.jit.trace(get_sam_decoder()_model, [torch.tensor(data[0]) for _, data in get_sam_decoder()_sample_inputs.items()])
166
 
167
  # Compile model on a specific device
168
- get_sam_decoder()_compile_job = hub.submit_compile_job(
169
- model=traced_get_sam_decoder()_model ,
170
  device=device,
171
- input_specs=get_sam_decoder()_model.get_input_spec(),
172
  )
173
 
174
  # Get target model to run on-device
175
- get_sam_decoder()_target_model = get_sam_decoder()_compile_job.get_target_model()
176
  # Trace model
177
- get_sam_encoder()_input_shape = get_sam_encoder()_model.get_input_spec()
178
- get_sam_encoder()_sample_inputs = get_sam_encoder()_model.sample_inputs()
179
 
180
- traced_get_sam_encoder()_model = torch.jit.trace(get_sam_encoder()_model, [torch.tensor(data[0]) for _, data in get_sam_encoder()_sample_inputs.items()])
181
 
182
  # Compile model on a specific device
183
- get_sam_encoder()_compile_job = hub.submit_compile_job(
184
- model=traced_get_sam_encoder()_model ,
185
  device=device,
186
- input_specs=get_sam_encoder()_model.get_input_spec(),
187
  )
188
 
189
  # Get target model to run on-device
190
- get_sam_encoder()_target_model = get_sam_encoder()_compile_job.get_target_model()
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
191
 
192
  ```
193
 
@@ -199,12 +450,32 @@ After compiling models from step 1. Models can be profiled model on-device using
199
  provisioned in the cloud. Once the job is submitted, you can navigate to a
200
  provided job URL to view a variety of on-device performance metrics.
201
  ```python
202
- get_sam_decoder()_profile_job = hub.submit_profile_job(
203
- model=get_sam_decoder()_target_model,
204
  device=device,
205
  )
206
- get_sam_encoder()_profile_job = hub.submit_profile_job(
207
- model=get_sam_encoder()_target_model,
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
208
  device=device,
209
  )
210
 
@@ -215,20 +486,55 @@ Step 3: **Verify on-device accuracy**
215
  To verify the accuracy of the model on-device, you can run on-device inference
216
  on sample input data on the same cloud hosted device.
217
  ```python
218
- get_sam_decoder()_input_data = get_sam_decoder()_model.sample_inputs()
219
- get_sam_decoder()_inference_job = hub.submit_inference_job(
220
- model=get_sam_decoder()_target_model,
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
221
  device=device,
222
- inputs=get_sam_decoder()_input_data,
223
  )
224
- get_sam_decoder()_inference_job.download_output_data()
225
- get_sam_encoder()_input_data = get_sam_encoder()_model.sample_inputs()
226
- get_sam_encoder()_inference_job = hub.submit_inference_job(
227
- model=get_sam_encoder()_target_model,
228
  device=device,
229
- inputs=get_sam_encoder()_input_data,
230
  )
231
- get_sam_encoder()_inference_job.download_output_data()
232
 
233
  ```
234
  With the output of the model, you can compute like PSNR, relative errors or
 
35
 
36
  | Model | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Precision | Primary Compute Unit | Target Model
37
  |---|---|---|---|---|---|---|---|---|
38
+ | SAMDecoder | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | TFLITE | 7.557 ms | 0 - 31 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMDecoder.tflite) |
39
+ | SAMDecoder | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | QNN | 7.052 ms | 4 - 19 MB | FP16 | NPU | [Segment-Anything-Model.so](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMDecoder.so) |
40
+ | SAMDecoder | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | ONNX | 11.415 ms | 0 - 266 MB | FP16 | NPU | [Segment-Anything-Model.onnx](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMDecoder.onnx) |
41
+ | SAMDecoder | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | TFLITE | 5.23 ms | 0 - 39 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMDecoder.tflite) |
42
+ | SAMDecoder | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | QNN | 4.973 ms | 4 - 46 MB | FP16 | NPU | [Segment-Anything-Model.so](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMDecoder.so) |
43
+ | SAMDecoder | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | ONNX | 7.741 ms | 5 - 126 MB | FP16 | NPU | [Segment-Anything-Model.onnx](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMDecoder.onnx) |
44
+ | SAMDecoder | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | TFLITE | 5.18 ms | 0 - 35 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMDecoder.tflite) |
45
+ | SAMDecoder | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | QNN | 3.787 ms | 4 - 42 MB | FP16 | NPU | Use Export Script |
46
+ | SAMDecoder | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | ONNX | 8.231 ms | 0 - 86 MB | FP16 | NPU | [Segment-Anything-Model.onnx](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMDecoder.onnx) |
47
+ | SAMDecoder | QCS8550 (Proxy) | QCS8550 Proxy | TFLITE | 7.484 ms | 0 - 33 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMDecoder.tflite) |
48
+ | SAMDecoder | QCS8550 (Proxy) | QCS8550 Proxy | QNN | 6.814 ms | 4 - 5 MB | FP16 | NPU | Use Export Script |
49
+ | SAMDecoder | SA7255P ADP | SA7255P | TFLITE | 52.929 ms | 0 - 32 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMDecoder.tflite) |
50
+ | SAMDecoder | SA7255P ADP | SA7255P | QNN | 50.031 ms | 2 - 11 MB | FP16 | NPU | Use Export Script |
51
+ | SAMDecoder | SA8255 (Proxy) | SA8255P Proxy | TFLITE | 7.484 ms | 0 - 32 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMDecoder.tflite) |
52
+ | SAMDecoder | SA8255 (Proxy) | SA8255P Proxy | QNN | 6.852 ms | 4 - 5 MB | FP16 | NPU | Use Export Script |
53
+ | SAMDecoder | SA8295P ADP | SA8295P | TFLITE | 9.976 ms | 0 - 35 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMDecoder.tflite) |
54
+ | SAMDecoder | SA8295P ADP | SA8295P | QNN | 9.05 ms | 0 - 6 MB | FP16 | NPU | Use Export Script |
55
+ | SAMDecoder | SA8650 (Proxy) | SA8650P Proxy | TFLITE | 7.477 ms | 0 - 29 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMDecoder.tflite) |
56
+ | SAMDecoder | SA8650 (Proxy) | SA8650P Proxy | QNN | 6.893 ms | 4 - 6 MB | FP16 | NPU | Use Export Script |
57
+ | SAMDecoder | SA8775P ADP | SA8775P | TFLITE | 10.51 ms | 0 - 34 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMDecoder.tflite) |
58
+ | SAMDecoder | SA8775P ADP | SA8775P | QNN | 9.667 ms | 0 - 6 MB | FP16 | NPU | Use Export Script |
59
+ | SAMDecoder | QCS8450 (Proxy) | QCS8450 Proxy | TFLITE | 8.533 ms | 0 - 37 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMDecoder.tflite) |
60
+ | SAMDecoder | QCS8450 (Proxy) | QCS8450 Proxy | QNN | 8.315 ms | 4 - 44 MB | FP16 | NPU | Use Export Script |
61
+ | SAMDecoder | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN | 7.458 ms | 4 - 4 MB | FP16 | NPU | Use Export Script |
62
+ | SAMDecoder | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 14.75 ms | 12 - 12 MB | FP16 | NPU | [Segment-Anything-Model.onnx](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMDecoder.onnx) |
63
+ | SAMEncoderPart1 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | TFLITE | 208.201 ms | 12 - 74 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart1.tflite) |
64
+ | SAMEncoderPart1 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | QNN | 203.885 ms | 12 - 67 MB | FP16 | NPU | [Segment-Anything-Model.so](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart1.so) |
65
+ | SAMEncoderPart1 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | ONNX | 168.865 ms | 12 - 57 MB | FP16 | NPU | [Segment-Anything-Model.onnx](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart1.onnx) |
66
+ | SAMEncoderPart1 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | TFLITE | 149.759 ms | 11 - 660 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart1.tflite) |
67
+ | SAMEncoderPart1 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | QNN | 144.476 ms | 364 - 1006 MB | FP16 | NPU | [Segment-Anything-Model.so](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart1.so) |
68
+ | SAMEncoderPart1 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | ONNX | 121.68 ms | 0 - 1234 MB | FP16 | NPU | [Segment-Anything-Model.onnx](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart1.onnx) |
69
+ | SAMEncoderPart1 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | TFLITE | 128.622 ms | 11 - 667 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart1.tflite) |
70
+ | SAMEncoderPart1 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | QNN | 141.432 ms | 3 - 655 MB | FP16 | NPU | Use Export Script |
71
+ | SAMEncoderPart1 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | ONNX | 96.596 ms | 39 - 1086 MB | FP16 | NPU | [Segment-Anything-Model.onnx](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart1.onnx) |
72
+ | SAMEncoderPart1 | QCS8550 (Proxy) | QCS8550 Proxy | TFLITE | 205.315 ms | 12 - 81 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart1.tflite) |
73
+ | SAMEncoderPart1 | QCS8550 (Proxy) | QCS8550 Proxy | QNN | 177.157 ms | 12 - 14 MB | FP16 | NPU | Use Export Script |
74
+ | SAMEncoderPart1 | SA7255P ADP | SA7255P | TFLITE | 1173.558 ms | 2 - 646 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart1.tflite) |
75
+ | SAMEncoderPart1 | SA7255P ADP | SA7255P | QNN | 1100.71 ms | 4 - 13 MB | FP16 | NPU | Use Export Script |
76
+ | SAMEncoderPart1 | SA8255 (Proxy) | SA8255P Proxy | TFLITE | 208.837 ms | 12 - 70 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart1.tflite) |
77
+ | SAMEncoderPart1 | SA8255 (Proxy) | SA8255P Proxy | QNN | 178.44 ms | 12 - 14 MB | FP16 | NPU | Use Export Script |
78
+ | SAMEncoderPart1 | SA8295P ADP | SA8295P | TFLITE | 242.752 ms | 12 - 640 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart1.tflite) |
79
+ | SAMEncoderPart1 | SA8295P ADP | SA8295P | QNN | 206.972 ms | 0 - 6 MB | FP16 | NPU | Use Export Script |
80
+ | SAMEncoderPart1 | SA8650 (Proxy) | SA8650P Proxy | TFLITE | 208.397 ms | 12 - 65 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart1.tflite) |
81
+ | SAMEncoderPart1 | SA8650 (Proxy) | SA8650P Proxy | QNN | 177.673 ms | 13 - 14 MB | FP16 | NPU | Use Export Script |
82
+ | SAMEncoderPart1 | SA8775P ADP | SA8775P | TFLITE | 251.286 ms | 12 - 656 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart1.tflite) |
83
+ | SAMEncoderPart1 | SA8775P ADP | SA8775P | QNN | 211.825 ms | 3 - 9 MB | FP16 | NPU | Use Export Script |
84
+ | SAMEncoderPart1 | QCS8450 (Proxy) | QCS8450 Proxy | TFLITE | 226.498 ms | 12 - 995 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart1.tflite) |
85
+ | SAMEncoderPart1 | QCS8450 (Proxy) | QCS8450 Proxy | QNN | 219.929 ms | 9 - 964 MB | FP16 | NPU | Use Export Script |
86
+ | SAMEncoderPart1 | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN | 171.827 ms | 12 - 12 MB | FP16 | NPU | Use Export Script |
87
+ | SAMEncoderPart1 | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 179.691 ms | 39 - 39 MB | FP16 | NPU | [Segment-Anything-Model.onnx](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart1.onnx) |
88
+ | SAMEncoderPart2 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | TFLITE | 670.133 ms | 12 - 104 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart2.tflite) |
89
+ | SAMEncoderPart2 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | QNN | 838.379 ms | 12 - 109 MB | FP16 | NPU | [Segment-Anything-Model.so](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart2.so) |
90
+ | SAMEncoderPart2 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | ONNX | 722.337 ms | 0 - 54 MB | FP16 | NPU | [Segment-Anything-Model.onnx](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart2.onnx) |
91
+ | SAMEncoderPart2 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | TFLITE | 556.374 ms | 11 - 1092 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart2.tflite) |
92
+ | SAMEncoderPart2 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | QNN | 668.897 ms | 12 - 1113 MB | FP16 | NPU | [Segment-Anything-Model.so](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart2.so) |
93
+ | SAMEncoderPart2 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | TFLITE | 487.43 ms | 12 - 1100 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart2.tflite) |
94
+ | SAMEncoderPart2 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | QNN | 574.723 ms | 10 - 1113 MB | FP16 | NPU | Use Export Script |
95
+ | SAMEncoderPart2 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | ONNX | 474.562 ms | 41 - 4748 MB | FP16 | NPU | [Segment-Anything-Model.onnx](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart2.onnx) |
96
+ | SAMEncoderPart2 | QCS8550 (Proxy) | QCS8550 Proxy | TFLITE | 671.625 ms | 0 - 102 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart2.tflite) |
97
+ | SAMEncoderPart2 | QCS8550 (Proxy) | QCS8550 Proxy | QNN | 737.788 ms | 13 - 14 MB | FP16 | NPU | Use Export Script |
98
+ | SAMEncoderPart2 | SA7255P ADP | SA7255P | QNN | 1869.513 ms | 0 - 10 MB | FP16 | NPU | Use Export Script |
99
+ | SAMEncoderPart2 | SA8255 (Proxy) | SA8255P Proxy | TFLITE | 690.907 ms | 12 - 116 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart2.tflite) |
100
+ | SAMEncoderPart2 | SA8255 (Proxy) | SA8255P Proxy | QNN | 731.994 ms | 15 - 16 MB | FP16 | NPU | Use Export Script |
101
+ | SAMEncoderPart2 | SA8295P ADP | SA8295P | TFLITE | 726.454 ms | 12 - 1141 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart2.tflite) |
102
+ | SAMEncoderPart2 | SA8295P ADP | SA8295P | QNN | 782.399 ms | 0 - 6 MB | FP16 | NPU | Use Export Script |
103
+ | SAMEncoderPart2 | SA8650 (Proxy) | SA8650P Proxy | TFLITE | 671.719 ms | 12 - 104 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart2.tflite) |
104
+ | SAMEncoderPart2 | SA8650 (Proxy) | SA8650P Proxy | QNN | 733.997 ms | 12 - 14 MB | FP16 | NPU | Use Export Script |
105
+ | SAMEncoderPart2 | SA8775P ADP | SA8775P | TFLITE | 717.157 ms | 0 - 1105 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart2.tflite) |
106
+ | SAMEncoderPart2 | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN | 633.76 ms | 12 - 12 MB | FP16 | NPU | Use Export Script |
107
+ | SAMEncoderPart2 | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 743.453 ms | 51 - 51 MB | FP16 | NPU | [Segment-Anything-Model.onnx](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart2.onnx) |
108
+ | SAMEncoderPart3 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | TFLITE | 686.379 ms | 12 - 103 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart3.tflite) |
109
+ | SAMEncoderPart3 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | QNN | 852.964 ms | 12 - 115 MB | FP16 | NPU | [Segment-Anything-Model.so](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart3.so) |
110
+ | SAMEncoderPart3 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | ONNX | 740.438 ms | 9 - 62 MB | FP16 | NPU | [Segment-Anything-Model.onnx](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart3.onnx) |
111
+ | SAMEncoderPart3 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | TFLITE | 440.773 ms | 12 - 1099 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart3.tflite) |
112
+ | SAMEncoderPart3 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | QNN | 579.622 ms | 11 - 1115 MB | FP16 | NPU | Use Export Script |
113
+ | SAMEncoderPart3 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | ONNX | 472.649 ms | 39 - 4748 MB | FP16 | NPU | [Segment-Anything-Model.onnx](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart3.onnx) |
114
+ | SAMEncoderPart3 | QCS8550 (Proxy) | QCS8550 Proxy | TFLITE | 668.273 ms | 12 - 114 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart3.tflite) |
115
+ | SAMEncoderPart3 | QCS8550 (Proxy) | QCS8550 Proxy | QNN | 731.938 ms | 12 - 14 MB | FP16 | NPU | Use Export Script |
116
+ | SAMEncoderPart3 | SA7255P ADP | SA7255P | QNN | 1874.534 ms | 12 - 22 MB | FP16 | NPU | Use Export Script |
117
+ | SAMEncoderPart3 | SA8255 (Proxy) | SA8255P Proxy | TFLITE | 684.379 ms | 12 - 108 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart3.tflite) |
118
+ | SAMEncoderPart3 | SA8255 (Proxy) | SA8255P Proxy | QNN | 731.638 ms | 13 - 15 MB | FP16 | NPU | Use Export Script |
119
+ | SAMEncoderPart3 | SA8295P ADP | SA8295P | TFLITE | 725.387 ms | 12 - 1148 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart3.tflite) |
120
+ | SAMEncoderPart3 | SA8295P ADP | SA8295P | QNN | 781.146 ms | 0 - 6 MB | FP16 | NPU | Use Export Script |
121
+ | SAMEncoderPart3 | SA8650 (Proxy) | SA8650P Proxy | TFLITE | 683.645 ms | 12 - 112 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart3.tflite) |
122
+ | SAMEncoderPart3 | SA8650 (Proxy) | SA8650P Proxy | QNN | 735.421 ms | 13 - 14 MB | FP16 | NPU | Use Export Script |
123
+ | SAMEncoderPart3 | SA8775P ADP | SA8775P | TFLITE | 724.877 ms | 0 - 1107 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart3.tflite) |
124
+ | SAMEncoderPart3 | SA8775P ADP | SA8775P | QNN | 742.231 ms | 1 - 11 MB | FP16 | NPU | Use Export Script |
125
+ | SAMEncoderPart3 | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN | 632.458 ms | 12 - 12 MB | FP16 | NPU | Use Export Script |
126
+ | SAMEncoderPart3 | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 741.655 ms | 51 - 51 MB | FP16 | NPU | [Segment-Anything-Model.onnx](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart3.onnx) |
127
+ | SAMEncoderPart4 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | QNN | 833.428 ms | 12 - 100 MB | FP16 | NPU | [Segment-Anything-Model.so](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart4.so) |
128
+ | SAMEncoderPart4 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | ONNX | 682.163 ms | 0 - 53 MB | FP16 | NPU | [Segment-Anything-Model.onnx](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart4.onnx) |
129
+ | SAMEncoderPart4 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | TFLITE | 437.593 ms | 11 - 1098 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart4.tflite) |
130
+ | SAMEncoderPart4 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | QNN | 531.361 ms | 10 - 1119 MB | FP16 | NPU | Use Export Script |
131
+ | SAMEncoderPart4 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | ONNX | 563.94 ms | 26 - 5204 MB | FP16 | NPU | [Segment-Anything-Model.onnx](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart4.onnx) |
132
+ | SAMEncoderPart4 | QCS8550 (Proxy) | QCS8550 Proxy | TFLITE | 675.986 ms | 12 - 102 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart4.tflite) |
133
+ | SAMEncoderPart4 | QCS8550 (Proxy) | QCS8550 Proxy | QNN | 731.308 ms | 12 - 14 MB | FP16 | NPU | Use Export Script |
134
+ | SAMEncoderPart4 | SA8255 (Proxy) | SA8255P Proxy | TFLITE | 677.387 ms | 12 - 103 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart4.tflite) |
135
+ | SAMEncoderPart4 | SA8255 (Proxy) | SA8255P Proxy | QNN | 740.258 ms | 13 - 14 MB | FP16 | NPU | Use Export Script |
136
+ | SAMEncoderPart4 | SA8295P ADP | SA8295P | TFLITE | 725.343 ms | 12 - 1150 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart4.tflite) |
137
+ | SAMEncoderPart4 | SA8295P ADP | SA8295P | QNN | 782.501 ms | 0 - 6 MB | FP16 | NPU | Use Export Script |
138
+ | SAMEncoderPart4 | SA8650 (Proxy) | SA8650P Proxy | TFLITE | 674.544 ms | 12 - 102 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart4.tflite) |
139
+ | SAMEncoderPart4 | SA8650 (Proxy) | SA8650P Proxy | QNN | 745.54 ms | 13 - 14 MB | FP16 | NPU | Use Export Script |
140
+ | SAMEncoderPart4 | SA8775P ADP | SA8775P | QNN | 738.7 ms | 12 - 22 MB | FP16 | NPU | Use Export Script |
141
+ | SAMEncoderPart4 | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN | 628.108 ms | 12 - 12 MB | FP16 | NPU | Use Export Script |
142
+ | SAMEncoderPart4 | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 739.45 ms | 53 - 53 MB | FP16 | NPU | [Segment-Anything-Model.onnx](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart4.onnx) |
143
+ | SAMEncoderPart5 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | TFLITE | 671.428 ms | 12 - 107 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart5.tflite) |
144
+ | SAMEncoderPart5 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | QNN | 844.265 ms | 12 - 102 MB | FP16 | NPU | [Segment-Anything-Model.so](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart5.so) |
145
+ | SAMEncoderPart5 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | ONNX | 684.639 ms | 0 - 53 MB | FP16 | NPU | [Segment-Anything-Model.onnx](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart5.onnx) |
146
+ | SAMEncoderPart5 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | TFLITE | 548.457 ms | 12 - 1096 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart5.tflite) |
147
+ | SAMEncoderPart5 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | TFLITE | 484.471 ms | 11 - 1099 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart5.tflite) |
148
+ | SAMEncoderPart5 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | QNN | 575.509 ms | 7 - 1111 MB | FP16 | NPU | Use Export Script |
149
+ | SAMEncoderPart5 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | ONNX | 470.838 ms | 39 - 4742 MB | FP16 | NPU | [Segment-Anything-Model.onnx](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart5.onnx) |
150
+ | SAMEncoderPart5 | QCS8550 (Proxy) | QCS8550 Proxy | TFLITE | 665.458 ms | 12 - 104 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart5.tflite) |
151
+ | SAMEncoderPart5 | QCS8550 (Proxy) | QCS8550 Proxy | QNN | 720.635 ms | 12 - 14 MB | FP16 | NPU | Use Export Script |
152
+ | SAMEncoderPart5 | SA7255P ADP | SA7255P | QNN | 1874.681 ms | 11 - 22 MB | FP16 | NPU | Use Export Script |
153
+ | SAMEncoderPart5 | SA8255 (Proxy) | SA8255P Proxy | TFLITE | 691.758 ms | 12 - 113 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart5.tflite) |
154
+ | SAMEncoderPart5 | SA8255 (Proxy) | SA8255P Proxy | QNN | 745.957 ms | 13 - 14 MB | FP16 | NPU | Use Export Script |
155
+ | SAMEncoderPart5 | SA8295P ADP | SA8295P | TFLITE | 725.701 ms | 12 - 1149 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart5.tflite) |
156
+ | SAMEncoderPart5 | SA8295P ADP | SA8295P | QNN | 782.525 ms | 0 - 6 MB | FP16 | NPU | Use Export Script |
157
+ | SAMEncoderPart5 | SA8650 (Proxy) | SA8650P Proxy | TFLITE | 679.545 ms | 12 - 105 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart5.tflite) |
158
+ | SAMEncoderPart5 | SA8650 (Proxy) | SA8650P Proxy | QNN | 730.129 ms | 13 - 15 MB | FP16 | NPU | Use Export Script |
159
+ | SAMEncoderPart5 | SA8775P ADP | SA8775P | QNN | 741.02 ms | 1 - 11 MB | FP16 | NPU | Use Export Script |
160
+ | SAMEncoderPart5 | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN | 640.949 ms | 12 - 12 MB | FP16 | NPU | Use Export Script |
161
+ | SAMEncoderPart5 | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 735.548 ms | 51 - 51 MB | FP16 | NPU | [Segment-Anything-Model.onnx](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart5.onnx) |
162
+ | SAMEncoderPart6 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | TFLITE | 677.701 ms | 12 - 102 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart6.tflite) |
163
+ | SAMEncoderPart6 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | QNN | 841.659 ms | 12 - 109 MB | FP16 | NPU | [Segment-Anything-Model.so](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart6.so) |
164
+ | SAMEncoderPart6 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | ONNX | 684.635 ms | 0 - 935 MB | FP16 | NPU | [Segment-Anything-Model.onnx](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart6.onnx) |
165
+ | SAMEncoderPart6 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | TFLITE | 536.637 ms | 10 - 1099 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart6.tflite) |
166
+ | SAMEncoderPart6 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | TFLITE | 485.849 ms | 11 - 1093 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart6.tflite) |
167
+ | SAMEncoderPart6 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | QNN | 579.641 ms | 10 - 1114 MB | FP16 | NPU | Use Export Script |
168
+ | SAMEncoderPart6 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | ONNX | 472.01 ms | 39 - 4745 MB | FP16 | NPU | [Segment-Anything-Model.onnx](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart6.onnx) |
169
+ | SAMEncoderPart6 | QCS8550 (Proxy) | QCS8550 Proxy | TFLITE | 656.062 ms | 12 - 113 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart6.tflite) |
170
+ | SAMEncoderPart6 | QCS8550 (Proxy) | QCS8550 Proxy | QNN | 730.888 ms | 12 - 14 MB | FP16 | NPU | Use Export Script |
171
+ | SAMEncoderPart6 | SA7255P ADP | SA7255P | QNN | 1868.576 ms | 5 - 15 MB | FP16 | NPU | Use Export Script |
172
+ | SAMEncoderPart6 | SA8255 (Proxy) | SA8255P Proxy | TFLITE | 688.443 ms | 12 - 102 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart6.tflite) |
173
+ | SAMEncoderPart6 | SA8255 (Proxy) | SA8255P Proxy | QNN | 736.226 ms | 12 - 14 MB | FP16 | NPU | Use Export Script |
174
+ | SAMEncoderPart6 | SA8295P ADP | SA8295P | TFLITE | 725.807 ms | 5 - 1142 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart6.tflite) |
175
+ | SAMEncoderPart6 | SA8295P ADP | SA8295P | QNN | 782.131 ms | 0 - 6 MB | FP16 | NPU | Use Export Script |
176
+ | SAMEncoderPart6 | SA8650 (Proxy) | SA8650P Proxy | TFLITE | 672.876 ms | 12 - 104 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart6.tflite) |
177
+ | SAMEncoderPart6 | SA8650 (Proxy) | SA8650P Proxy | QNN | 740.495 ms | 16 - 18 MB | FP16 | NPU | Use Export Script |
178
+ | SAMEncoderPart6 | SA8775P ADP | SA8775P | TFLITE | 726.536 ms | 0 - 1104 MB | FP16 | NPU | [Segment-Anything-Model.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart6.tflite) |
179
+ | SAMEncoderPart6 | SA8775P ADP | SA8775P | QNN | 742.038 ms | 2 - 12 MB | FP16 | NPU | Use Export Script |
180
+ | SAMEncoderPart6 | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN | 629.137 ms | 12 - 12 MB | FP16 | NPU | Use Export Script |
181
+ | SAMEncoderPart6 | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 743.947 ms | 52 - 52 MB | FP16 | NPU | [Segment-Anything-Model.onnx](https://huggingface.co/qualcomm/Segment-Anything-Model/blob/main/SAMEncoderPart6.onnx) |
182
 
183
 
184
 
 
243
  SAMDecoder
244
  Device : Samsung Galaxy S23 (13)
245
  Runtime : TFLITE
246
+ Estimated inference time (ms) : 7.6
247
+ Estimated peak memory usage (MB): [0, 31]
248
+ Total # Ops : 845
249
+ Compute Unit(s) : NPU (845 ops)
250
 
251
  ------------------------------------------------------------
252
+ SAMEncoderPart1
253
+ Device : Samsung Galaxy S23 (13)
254
+ Runtime : TFLITE
255
+ Estimated inference time (ms) : 208.2
256
+ Estimated peak memory usage (MB): [12, 74]
257
+ Total # Ops : 585
258
+ Compute Unit(s) : NPU (585 ops)
259
+
260
+ ------------------------------------------------------------
261
+ SAMEncoderPart2
262
+ Device : Samsung Galaxy S23 (13)
263
+ Runtime : TFLITE
264
+ Estimated inference time (ms) : 670.1
265
+ Estimated peak memory usage (MB): [12, 104]
266
+ Total # Ops : 580
267
+ Compute Unit(s) : NPU (580 ops)
268
+
269
+ ------------------------------------------------------------
270
+ SAMEncoderPart3
271
+ Device : Samsung Galaxy S23 (13)
272
+ Runtime : TFLITE
273
+ Estimated inference time (ms) : 686.4
274
+ Estimated peak memory usage (MB): [12, 103]
275
+ Total # Ops : 580
276
+ Compute Unit(s) : NPU (580 ops)
277
+
278
+ ------------------------------------------------------------
279
+ SAMEncoderPart4
280
+ Device : Samsung Galaxy S23 (13)
281
+ Runtime : QNN
282
+ Estimated inference time (ms) : 833.4
283
+ Estimated peak memory usage (MB): [12, 100]
284
+ Total # Ops : 613
285
+ Compute Unit(s) : NPU (613 ops)
286
+
287
+ ------------------------------------------------------------
288
+ SAMEncoderPart5
289
+ Device : Samsung Galaxy S23 (13)
290
+ Runtime : TFLITE
291
+ Estimated inference time (ms) : 671.4
292
+ Estimated peak memory usage (MB): [12, 107]
293
+ Total # Ops : 580
294
+ Compute Unit(s) : NPU (580 ops)
295
+
296
+ ------------------------------------------------------------
297
+ SAMEncoderPart6
298
+ Device : Samsung Galaxy S23 (13)
299
+ Runtime : TFLITE
300
+ Estimated inference time (ms) : 677.7
301
+ Estimated peak memory usage (MB): [12, 102]
302
+ Total # Ops : 580
303
+ Compute Unit(s) : NPU (580 ops)
304
  ```
305
 
306
 
 
323
 
324
  # Load the model
325
  model = Model.from_pretrained()
326
+ decoder_model = model.decoder
327
+ encoder_splits[0]_model = model.encoder_splits[0]
328
+ encoder_splits[1]_model = model.encoder_splits[1]
329
+ encoder_splits[2]_model = model.encoder_splits[2]
330
+ encoder_splits[3]_model = model.encoder_splits[3]
331
+ encoder_splits[4]_model = model.encoder_splits[4]
332
+ encoder_splits[5]_model = model.encoder_splits[5]
333
 
334
  # Device
335
  device = hub.Device("Samsung Galaxy S23")
336
 
337
  # Trace model
338
+ decoder_input_shape = decoder_model.get_input_spec()
339
+ decoder_sample_inputs = decoder_model.sample_inputs()
340
 
341
+ traced_decoder_model = torch.jit.trace(decoder_model, [torch.tensor(data[0]) for _, data in decoder_sample_inputs.items()])
342
 
343
  # Compile model on a specific device
344
+ decoder_compile_job = hub.submit_compile_job(
345
+ model=traced_decoder_model ,
346
  device=device,
347
+ input_specs=decoder_model.get_input_spec(),
348
  )
349
 
350
  # Get target model to run on-device
351
+ decoder_target_model = decoder_compile_job.get_target_model()
352
  # Trace model
353
+ encoder_splits[0]_input_shape = encoder_splits[0]_model.get_input_spec()
354
+ encoder_splits[0]_sample_inputs = encoder_splits[0]_model.sample_inputs()
355
 
356
+ traced_encoder_splits[0]_model = torch.jit.trace(encoder_splits[0]_model, [torch.tensor(data[0]) for _, data in encoder_splits[0]_sample_inputs.items()])
357
 
358
  # Compile model on a specific device
359
+ encoder_splits[0]_compile_job = hub.submit_compile_job(
360
+ model=traced_encoder_splits[0]_model ,
361
  device=device,
362
+ input_specs=encoder_splits[0]_model.get_input_spec(),
363
  )
364
 
365
  # Get target model to run on-device
366
+ encoder_splits[0]_target_model = encoder_splits[0]_compile_job.get_target_model()
367
+ # Trace model
368
+ encoder_splits[1]_input_shape = encoder_splits[1]_model.get_input_spec()
369
+ encoder_splits[1]_sample_inputs = encoder_splits[1]_model.sample_inputs()
370
+
371
+ traced_encoder_splits[1]_model = torch.jit.trace(encoder_splits[1]_model, [torch.tensor(data[0]) for _, data in encoder_splits[1]_sample_inputs.items()])
372
+
373
+ # Compile model on a specific device
374
+ encoder_splits[1]_compile_job = hub.submit_compile_job(
375
+ model=traced_encoder_splits[1]_model ,
376
+ device=device,
377
+ input_specs=encoder_splits[1]_model.get_input_spec(),
378
+ )
379
+
380
+ # Get target model to run on-device
381
+ encoder_splits[1]_target_model = encoder_splits[1]_compile_job.get_target_model()
382
+ # Trace model
383
+ encoder_splits[2]_input_shape = encoder_splits[2]_model.get_input_spec()
384
+ encoder_splits[2]_sample_inputs = encoder_splits[2]_model.sample_inputs()
385
+
386
+ traced_encoder_splits[2]_model = torch.jit.trace(encoder_splits[2]_model, [torch.tensor(data[0]) for _, data in encoder_splits[2]_sample_inputs.items()])
387
+
388
+ # Compile model on a specific device
389
+ encoder_splits[2]_compile_job = hub.submit_compile_job(
390
+ model=traced_encoder_splits[2]_model ,
391
+ device=device,
392
+ input_specs=encoder_splits[2]_model.get_input_spec(),
393
+ )
394
+
395
+ # Get target model to run on-device
396
+ encoder_splits[2]_target_model = encoder_splits[2]_compile_job.get_target_model()
397
+ # Trace model
398
+ encoder_splits[3]_input_shape = encoder_splits[3]_model.get_input_spec()
399
+ encoder_splits[3]_sample_inputs = encoder_splits[3]_model.sample_inputs()
400
+
401
+ traced_encoder_splits[3]_model = torch.jit.trace(encoder_splits[3]_model, [torch.tensor(data[0]) for _, data in encoder_splits[3]_sample_inputs.items()])
402
+
403
+ # Compile model on a specific device
404
+ encoder_splits[3]_compile_job = hub.submit_compile_job(
405
+ model=traced_encoder_splits[3]_model ,
406
+ device=device,
407
+ input_specs=encoder_splits[3]_model.get_input_spec(),
408
+ )
409
+
410
+ # Get target model to run on-device
411
+ encoder_splits[3]_target_model = encoder_splits[3]_compile_job.get_target_model()
412
+ # Trace model
413
+ encoder_splits[4]_input_shape = encoder_splits[4]_model.get_input_spec()
414
+ encoder_splits[4]_sample_inputs = encoder_splits[4]_model.sample_inputs()
415
+
416
+ traced_encoder_splits[4]_model = torch.jit.trace(encoder_splits[4]_model, [torch.tensor(data[0]) for _, data in encoder_splits[4]_sample_inputs.items()])
417
+
418
+ # Compile model on a specific device
419
+ encoder_splits[4]_compile_job = hub.submit_compile_job(
420
+ model=traced_encoder_splits[4]_model ,
421
+ device=device,
422
+ input_specs=encoder_splits[4]_model.get_input_spec(),
423
+ )
424
+
425
+ # Get target model to run on-device
426
+ encoder_splits[4]_target_model = encoder_splits[4]_compile_job.get_target_model()
427
+ # Trace model
428
+ encoder_splits[5]_input_shape = encoder_splits[5]_model.get_input_spec()
429
+ encoder_splits[5]_sample_inputs = encoder_splits[5]_model.sample_inputs()
430
+
431
+ traced_encoder_splits[5]_model = torch.jit.trace(encoder_splits[5]_model, [torch.tensor(data[0]) for _, data in encoder_splits[5]_sample_inputs.items()])
432
+
433
+ # Compile model on a specific device
434
+ encoder_splits[5]_compile_job = hub.submit_compile_job(
435
+ model=traced_encoder_splits[5]_model ,
436
+ device=device,
437
+ input_specs=encoder_splits[5]_model.get_input_spec(),
438
+ )
439
+
440
+ # Get target model to run on-device
441
+ encoder_splits[5]_target_model = encoder_splits[5]_compile_job.get_target_model()
442
 
443
  ```
444
 
 
450
  provisioned in the cloud. Once the job is submitted, you can navigate to a
451
  provided job URL to view a variety of on-device performance metrics.
452
  ```python
453
+ decoder_profile_job = hub.submit_profile_job(
454
+ model=decoder_target_model,
455
  device=device,
456
  )
457
+ encoder_splits[0]_profile_job = hub.submit_profile_job(
458
+ model=encoder_splits[0]_target_model,
459
+ device=device,
460
+ )
461
+ encoder_splits[1]_profile_job = hub.submit_profile_job(
462
+ model=encoder_splits[1]_target_model,
463
+ device=device,
464
+ )
465
+ encoder_splits[2]_profile_job = hub.submit_profile_job(
466
+ model=encoder_splits[2]_target_model,
467
+ device=device,
468
+ )
469
+ encoder_splits[3]_profile_job = hub.submit_profile_job(
470
+ model=encoder_splits[3]_target_model,
471
+ device=device,
472
+ )
473
+ encoder_splits[4]_profile_job = hub.submit_profile_job(
474
+ model=encoder_splits[4]_target_model,
475
+ device=device,
476
+ )
477
+ encoder_splits[5]_profile_job = hub.submit_profile_job(
478
+ model=encoder_splits[5]_target_model,
479
  device=device,
480
  )
481
 
 
486
  To verify the accuracy of the model on-device, you can run on-device inference
487
  on sample input data on the same cloud hosted device.
488
  ```python
489
+ decoder_input_data = decoder_model.sample_inputs()
490
+ decoder_inference_job = hub.submit_inference_job(
491
+ model=decoder_target_model,
492
+ device=device,
493
+ inputs=decoder_input_data,
494
+ )
495
+ decoder_inference_job.download_output_data()
496
+ encoder_splits[0]_input_data = encoder_splits[0]_model.sample_inputs()
497
+ encoder_splits[0]_inference_job = hub.submit_inference_job(
498
+ model=encoder_splits[0]_target_model,
499
+ device=device,
500
+ inputs=encoder_splits[0]_input_data,
501
+ )
502
+ encoder_splits[0]_inference_job.download_output_data()
503
+ encoder_splits[1]_input_data = encoder_splits[1]_model.sample_inputs()
504
+ encoder_splits[1]_inference_job = hub.submit_inference_job(
505
+ model=encoder_splits[1]_target_model,
506
+ device=device,
507
+ inputs=encoder_splits[1]_input_data,
508
+ )
509
+ encoder_splits[1]_inference_job.download_output_data()
510
+ encoder_splits[2]_input_data = encoder_splits[2]_model.sample_inputs()
511
+ encoder_splits[2]_inference_job = hub.submit_inference_job(
512
+ model=encoder_splits[2]_target_model,
513
+ device=device,
514
+ inputs=encoder_splits[2]_input_data,
515
+ )
516
+ encoder_splits[2]_inference_job.download_output_data()
517
+ encoder_splits[3]_input_data = encoder_splits[3]_model.sample_inputs()
518
+ encoder_splits[3]_inference_job = hub.submit_inference_job(
519
+ model=encoder_splits[3]_target_model,
520
+ device=device,
521
+ inputs=encoder_splits[3]_input_data,
522
+ )
523
+ encoder_splits[3]_inference_job.download_output_data()
524
+ encoder_splits[4]_input_data = encoder_splits[4]_model.sample_inputs()
525
+ encoder_splits[4]_inference_job = hub.submit_inference_job(
526
+ model=encoder_splits[4]_target_model,
527
  device=device,
528
+ inputs=encoder_splits[4]_input_data,
529
  )
530
+ encoder_splits[4]_inference_job.download_output_data()
531
+ encoder_splits[5]_input_data = encoder_splits[5]_model.sample_inputs()
532
+ encoder_splits[5]_inference_job = hub.submit_inference_job(
533
+ model=encoder_splits[5]_target_model,
534
  device=device,
535
+ inputs=encoder_splits[5]_input_data,
536
  )
537
+ encoder_splits[5]_inference_job.download_output_data()
538
 
539
  ```
540
  With the output of the model, you can compute like PSNR, relative errors or