qaihm-bot commited on
Commit
4e03924
·
verified ·
1 Parent(s): 2f22497

See https://github.com/quic/ai-hub-models/releases/v0.38.0 for changelog.

.gitattributes CHANGED
@@ -36,3 +36,5 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
36
  Segment-Anything-Model-2_SAM2Decoder.dlc filter=lfs diff=lfs merge=lfs -text
37
  Segment-Anything-Model-2_SAM2Encoder.dlc filter=lfs diff=lfs merge=lfs -text
38
  DEPLOYMENT_MODEL_LICENSE.pdf filter=lfs diff=lfs merge=lfs -text
 
 
 
36
  Segment-Anything-Model-2_SAM2Decoder.dlc filter=lfs diff=lfs merge=lfs -text
37
  Segment-Anything-Model-2_SAM2Encoder.dlc filter=lfs diff=lfs merge=lfs -text
38
  DEPLOYMENT_MODEL_LICENSE.pdf filter=lfs diff=lfs merge=lfs -text
39
+ Segment-Anything-Model-2_SAM2Decoder_float.dlc filter=lfs diff=lfs merge=lfs -text
40
+ Segment-Anything-Model-2_SAM2Encoder_float.dlc filter=lfs diff=lfs merge=lfs -text
README.md CHANGED
@@ -38,30 +38,36 @@ More details on model performance across various devices, can be found
38
 
39
  | Model | Precision | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit | Target Model
40
  |---|---|---|---|---|---|---|---|---|
41
- | SAM2Encoder | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 514.65 ms | 16 - 608 MB | NPU | [Segment-Anything-Model-2.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.tflite) |
42
- | SAM2Encoder | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 310.088 ms | 16 - 688 MB | NPU | [Segment-Anything-Model-2.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.tflite) |
43
- | SAM2Encoder | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 231.66 ms | 16 - 93 MB | NPU | [Segment-Anything-Model-2.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.tflite) |
44
- | SAM2Encoder | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 245.107 ms | 16 - 608 MB | NPU | [Segment-Anything-Model-2.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.tflite) |
45
- | SAM2Encoder | float | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 514.65 ms | 16 - 608 MB | NPU | [Segment-Anything-Model-2.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.tflite) |
46
- | SAM2Encoder | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 227.622 ms | 16 - 93 MB | NPU | [Segment-Anything-Model-2.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.tflite) |
47
- | SAM2Encoder | float | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 332.868 ms | 16 - 595 MB | NPU | [Segment-Anything-Model-2.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.tflite) |
48
- | SAM2Encoder | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 231.48 ms | 0 - 78 MB | NPU | [Segment-Anything-Model-2.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.tflite) |
49
- | SAM2Encoder | float | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 245.107 ms | 16 - 608 MB | NPU | [Segment-Anything-Model-2.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.tflite) |
50
- | SAM2Encoder | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | TFLITE | 231.752 ms | 16 - 90 MB | NPU | [Segment-Anything-Model-2.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.tflite) |
51
- | SAM2Encoder | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 167.166 ms | 412 - 1006 MB | NPU | [Segment-Anything-Model-2.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.tflite) |
52
- | SAM2Encoder | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | TFLITE | 154.567 ms | 11 - 601 MB | NPU | [Segment-Anything-Model-2.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.tflite) |
53
- | SAM2Decoder | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 17.534 ms | 0 - 52 MB | NPU | [Segment-Anything-Model-2.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.tflite) |
54
- | SAM2Decoder | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 11.025 ms | 0 - 83 MB | NPU | [Segment-Anything-Model-2.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.tflite) |
55
- | SAM2Decoder | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 7.927 ms | 0 - 34 MB | NPU | [Segment-Anything-Model-2.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.tflite) |
56
- | SAM2Decoder | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 9.251 ms | 0 - 54 MB | NPU | [Segment-Anything-Model-2.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.tflite) |
57
- | SAM2Decoder | float | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 17.534 ms | 0 - 52 MB | NPU | [Segment-Anything-Model-2.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.tflite) |
58
- | SAM2Decoder | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 7.906 ms | 0 - 29 MB | NPU | [Segment-Anything-Model-2.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.tflite) |
59
- | SAM2Decoder | float | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 12.656 ms | 0 - 79 MB | NPU | [Segment-Anything-Model-2.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.tflite) |
60
- | SAM2Decoder | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 7.936 ms | 0 - 33 MB | NPU | [Segment-Anything-Model-2.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.tflite) |
61
- | SAM2Decoder | float | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 9.251 ms | 0 - 54 MB | NPU | [Segment-Anything-Model-2.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.tflite) |
62
- | SAM2Decoder | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | TFLITE | 7.953 ms | 0 - 30 MB | NPU | [Segment-Anything-Model-2.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.tflite) |
63
- | SAM2Decoder | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 5.443 ms | 0 - 61 MB | NPU | [Segment-Anything-Model-2.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.tflite) |
64
- | SAM2Decoder | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | TFLITE | 5.492 ms | 0 - 52 MB | NPU | [Segment-Anything-Model-2.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.tflite) |
 
 
 
 
 
 
65
 
66
 
67
 
@@ -143,7 +149,7 @@ from qai_hub_models.models.sam2 import Model
143
  torch_model = Model.from_pretrained()
144
 
145
  # Device
146
- device = hub.Device("Samsung Galaxy S24")
147
 
148
  # Trace model
149
  input_shape = torch_model.get_input_spec()
 
38
 
39
  | Model | Precision | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit | Target Model
40
  |---|---|---|---|---|---|---|---|---|
41
+ | SAM2Encoder | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 514.682 ms | 16 - 604 MB | NPU | [Segment-Anything-Model-2.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.tflite) |
42
+ | SAM2Encoder | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 307.679 ms | 16 - 685 MB | NPU | [Segment-Anything-Model-2.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.tflite) |
43
+ | SAM2Encoder | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 378.993 ms | 12 - 2666 MB | NPU | [Segment-Anything-Model-2.dlc](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.dlc) |
44
+ | SAM2Encoder | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 227.46 ms | 16 - 96 MB | NPU | [Segment-Anything-Model-2.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.tflite) |
45
+ | SAM2Encoder | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 198.101 ms | 12 - 84 MB | NPU | [Segment-Anything-Model-2.dlc](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.dlc) |
46
+ | SAM2Encoder | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 244.284 ms | 16 - 607 MB | NPU | [Segment-Anything-Model-2.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.tflite) |
47
+ | SAM2Encoder | float | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 514.682 ms | 16 - 604 MB | NPU | [Segment-Anything-Model-2.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.tflite) |
48
+ | SAM2Encoder | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 230.937 ms | 0 - 71 MB | NPU | [Segment-Anything-Model-2.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.tflite) |
49
+ | SAM2Encoder | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 199.417 ms | 12 - 87 MB | NPU | [Segment-Anything-Model-2.dlc](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.dlc) |
50
+ | SAM2Encoder | float | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 334.416 ms | 14 - 594 MB | NPU | [Segment-Anything-Model-2.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.tflite) |
51
+ | SAM2Encoder | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 231.375 ms | 16 - 97 MB | NPU | [Segment-Anything-Model-2.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.tflite) |
52
+ | SAM2Encoder | float | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 244.284 ms | 16 - 607 MB | NPU | [Segment-Anything-Model-2.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.tflite) |
53
+ | SAM2Encoder | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 165.997 ms | 14 - 605 MB | NPU | [Segment-Anything-Model-2.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.tflite) |
54
+ | SAM2Encoder | float | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | TFLITE | 132.253 ms | 14 - 603 MB | NPU | [Segment-Anything-Model-2.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.tflite) |
55
+ | SAM2Encoder | float | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 194.387 ms | 243 - 243 MB | NPU | [Segment-Anything-Model-2.dlc](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.dlc) |
56
+ | SAM2Decoder | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 17.859 ms | 0 - 51 MB | NPU | [Segment-Anything-Model-2.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.tflite) |
57
+ | SAM2Decoder | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 11.071 ms | 0 - 49 MB | NPU | [Segment-Anything-Model-2.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.tflite) |
58
+ | SAM2Decoder | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 13.191 ms | 16 - 86 MB | NPU | [Segment-Anything-Model-2.dlc](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.dlc) |
59
+ | SAM2Decoder | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 8.088 ms | 0 - 32 MB | NPU | [Segment-Anything-Model-2.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.tflite) |
60
+ | SAM2Decoder | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 6.899 ms | 16 - 36 MB | NPU | [Segment-Anything-Model-2.dlc](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.dlc) |
61
+ | SAM2Decoder | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 9.446 ms | 0 - 50 MB | NPU | [Segment-Anything-Model-2.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.tflite) |
62
+ | SAM2Decoder | float | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 17.859 ms | 0 - 51 MB | NPU | [Segment-Anything-Model-2.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.tflite) |
63
+ | SAM2Decoder | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 8.092 ms | 0 - 33 MB | NPU | [Segment-Anything-Model-2.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.tflite) |
64
+ | SAM2Decoder | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 6.835 ms | 15 - 34 MB | NPU | [Segment-Anything-Model-2.dlc](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.dlc) |
65
+ | SAM2Decoder | float | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 12.708 ms | 0 - 50 MB | NPU | [Segment-Anything-Model-2.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.tflite) |
66
+ | SAM2Decoder | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 8.02 ms | 0 - 34 MB | NPU | [Segment-Anything-Model-2.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.tflite) |
67
+ | SAM2Decoder | float | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 9.446 ms | 0 - 50 MB | NPU | [Segment-Anything-Model-2.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.tflite) |
68
+ | SAM2Decoder | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 5.525 ms | 0 - 59 MB | NPU | [Segment-Anything-Model-2.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.tflite) |
69
+ | SAM2Decoder | float | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | TFLITE | 4.775 ms | 0 - 55 MB | NPU | [Segment-Anything-Model-2.tflite](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.tflite) |
70
+ | SAM2Decoder | float | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 7.172 ms | 28 - 28 MB | NPU | [Segment-Anything-Model-2.dlc](https://huggingface.co/qualcomm/Segment-Anything-Model-2/blob/main/Segment-Anything-Model-2.dlc) |
71
 
72
 
73
 
 
149
  torch_model = Model.from_pretrained()
150
 
151
  # Device
152
+ device = hub.Device("Samsung Galaxy S25")
153
 
154
  # Trace model
155
  input_shape = torch_model.get_input_spec()
Segment-Anything-Model-2_SAM2Decoder_float.dlc ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:77c503250e0b5e74cc2c1163dc8d1b8f10cfbfa36aa0b89c3728f97bb895ca0f
3
+ size 25348980
Segment-Anything-Model-2_SAM2Encoder_float.dlc ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a395e44d15d04b4f499187d8fa81c84c974fca3ed763a787c7adcb125973eb5f
3
+ size 134443236
tool-versions.yaml CHANGED
@@ -1,4 +1,3 @@
1
  tool_versions:
2
- tflite:
3
- qairt: 2.37.0.250724175447_124859
4
- tflite: 2.17.0
 
1
  tool_versions:
2
+ qnn_dlc:
3
+ qairt: 2.38.0.250901140452_125126