qaihm-bot commited on
Commit
e223da8
·
verified ·
1 Parent(s): fc38591

See https://github.com/quic/ai-hub-models/releases/v0.34.0 for changelog.

README.md CHANGED
@@ -23,6 +23,7 @@ More details on model performance across various devices, can be found
23
  [here](https://aihub.qualcomm.com/models/fastsam_x).
24
 
25
 
 
26
  ### Model Details
27
 
28
  - **Model Type:** Model_use_case.semantic_segmentation
@@ -35,31 +36,31 @@ More details on model performance across various devices, can be found
35
 
36
  | Model | Precision | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit | Target Model
37
  |---|---|---|---|---|---|---|---|---|
38
- | FastSam-X | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 280.992 ms | 4 - 120 MB | NPU | [FastSam-X.tflite](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.tflite) |
39
  | FastSam-X | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 277.315 ms | 4 - 90 MB | NPU | [FastSam-X.dlc](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.dlc) |
40
- | FastSam-X | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 88.853 ms | 4 - 198 MB | NPU | [FastSam-X.tflite](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.tflite) |
41
  | FastSam-X | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 99.798 ms | 5 - 93 MB | NPU | [FastSam-X.dlc](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.dlc) |
42
- | FastSam-X | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 46.209 ms | 4 - 65 MB | NPU | [FastSam-X.tflite](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.tflite) |
43
  | FastSam-X | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 40.425 ms | 5 - 32 MB | NPU | [FastSam-X.dlc](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.dlc) |
44
- | FastSam-X | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 70.685 ms | 1 - 113 MB | NPU | [FastSam-X.tflite](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.tflite) |
45
  | FastSam-X | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 67.905 ms | 4 - 87 MB | NPU | [FastSam-X.dlc](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.dlc) |
46
- | FastSam-X | float | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 280.992 ms | 4 - 120 MB | NPU | [FastSam-X.tflite](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.tflite) |
47
  | FastSam-X | float | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 277.315 ms | 4 - 90 MB | NPU | [FastSam-X.dlc](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.dlc) |
48
- | FastSam-X | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 46.416 ms | 4 - 61 MB | NPU | [FastSam-X.tflite](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.tflite) |
49
  | FastSam-X | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 42.995 ms | 5 - 36 MB | NPU | [FastSam-X.dlc](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.dlc) |
50
- | FastSam-X | float | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 89.574 ms | 4 - 105 MB | NPU | [FastSam-X.tflite](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.tflite) |
51
  | FastSam-X | float | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 79.789 ms | 5 - 82 MB | NPU | [FastSam-X.dlc](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.dlc) |
52
- | FastSam-X | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 45.724 ms | 4 - 66 MB | NPU | [FastSam-X.tflite](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.tflite) |
53
  | FastSam-X | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_DLC | 42.818 ms | 5 - 30 MB | NPU | [FastSam-X.dlc](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.dlc) |
54
- | FastSam-X | float | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 70.685 ms | 1 - 113 MB | NPU | [FastSam-X.tflite](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.tflite) |
55
  | FastSam-X | float | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 67.905 ms | 4 - 87 MB | NPU | [FastSam-X.dlc](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.dlc) |
56
- | FastSam-X | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | TFLITE | 46.298 ms | 4 - 62 MB | NPU | [FastSam-X.tflite](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.tflite) |
57
  | FastSam-X | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | QNN_DLC | 42.361 ms | 5 - 30 MB | NPU | [FastSam-X.dlc](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.dlc) |
58
  | FastSam-X | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | ONNX | 46.52 ms | 0 - 174 MB | NPU | [FastSam-X.onnx](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.onnx) |
59
- | FastSam-X | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 35.006 ms | 3 - 210 MB | NPU | [FastSam-X.tflite](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.tflite) |
60
  | FastSam-X | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 32.931 ms | 5 - 86 MB | NPU | [FastSam-X.dlc](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.dlc) |
61
  | FastSam-X | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 35.033 ms | 14 - 98 MB | NPU | [FastSam-X.onnx](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.onnx) |
62
- | FastSam-X | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | TFLITE | 30.716 ms | 4 - 121 MB | NPU | [FastSam-X.tflite](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.tflite) |
63
  | FastSam-X | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | QNN_DLC | 25.394 ms | 5 - 91 MB | NPU | [FastSam-X.dlc](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.dlc) |
64
  | FastSam-X | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 29.851 ms | 14 - 88 MB | NPU | [FastSam-X.onnx](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.onnx) |
65
  | FastSam-X | float | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 45.562 ms | 355 - 355 MB | NPU | [FastSam-X.dlc](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.dlc) |
@@ -121,17 +122,7 @@ device. This script does the following:
121
  ```bash
122
  python -m qai_hub_models.models.fastsam_x.export
123
  ```
124
- ```
125
- Profiling Results
126
- ------------------------------------------------------------
127
- FastSam-X
128
- Device : cs_8275 (ANDROID 14)
129
- Runtime : TFLITE
130
- Estimated inference time (ms) : 281.0
131
- Estimated peak memory usage (MB): [4, 120]
132
- Total # Ops : 419
133
- Compute Unit(s) : npu (419 ops) gpu (0 ops) cpu (0 ops)
134
- ```
135
 
136
 
137
  ## How does this work?
 
23
  [here](https://aihub.qualcomm.com/models/fastsam_x).
24
 
25
 
26
+
27
  ### Model Details
28
 
29
  - **Model Type:** Model_use_case.semantic_segmentation
 
36
 
37
  | Model | Precision | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit | Target Model
38
  |---|---|---|---|---|---|---|---|---|
39
+ | FastSam-X | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 2097.531 ms | 0 - 115 MB | NPU | [FastSam-X.tflite](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.tflite) |
40
  | FastSam-X | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 277.315 ms | 4 - 90 MB | NPU | [FastSam-X.dlc](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.dlc) |
41
+ | FastSam-X | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 87.273 ms | 4 - 202 MB | NPU | [FastSam-X.tflite](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.tflite) |
42
  | FastSam-X | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 99.798 ms | 5 - 93 MB | NPU | [FastSam-X.dlc](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.dlc) |
43
+ | FastSam-X | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 43.436 ms | 4 - 62 MB | NPU | [FastSam-X.tflite](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.tflite) |
44
  | FastSam-X | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 40.425 ms | 5 - 32 MB | NPU | [FastSam-X.dlc](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.dlc) |
45
+ | FastSam-X | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 70.672 ms | 4 - 119 MB | NPU | [FastSam-X.tflite](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.tflite) |
46
  | FastSam-X | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 67.905 ms | 4 - 87 MB | NPU | [FastSam-X.dlc](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.dlc) |
47
+ | FastSam-X | float | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 2097.531 ms | 0 - 115 MB | NPU | [FastSam-X.tflite](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.tflite) |
48
  | FastSam-X | float | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 277.315 ms | 4 - 90 MB | NPU | [FastSam-X.dlc](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.dlc) |
49
+ | FastSam-X | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 45.819 ms | 4 - 62 MB | NPU | [FastSam-X.tflite](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.tflite) |
50
  | FastSam-X | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 42.995 ms | 5 - 36 MB | NPU | [FastSam-X.dlc](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.dlc) |
51
+ | FastSam-X | float | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 89.598 ms | 4 - 104 MB | NPU | [FastSam-X.tflite](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.tflite) |
52
  | FastSam-X | float | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 79.789 ms | 5 - 82 MB | NPU | [FastSam-X.dlc](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.dlc) |
53
+ | FastSam-X | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 43.576 ms | 4 - 63 MB | NPU | [FastSam-X.tflite](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.tflite) |
54
  | FastSam-X | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_DLC | 42.818 ms | 5 - 30 MB | NPU | [FastSam-X.dlc](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.dlc) |
55
+ | FastSam-X | float | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 70.672 ms | 4 - 119 MB | NPU | [FastSam-X.tflite](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.tflite) |
56
  | FastSam-X | float | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 67.905 ms | 4 - 87 MB | NPU | [FastSam-X.dlc](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.dlc) |
57
+ | FastSam-X | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | TFLITE | 45.814 ms | 0 - 63 MB | NPU | [FastSam-X.tflite](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.tflite) |
58
  | FastSam-X | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | QNN_DLC | 42.361 ms | 5 - 30 MB | NPU | [FastSam-X.dlc](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.dlc) |
59
  | FastSam-X | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | ONNX | 46.52 ms | 0 - 174 MB | NPU | [FastSam-X.onnx](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.onnx) |
60
+ | FastSam-X | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 34.452 ms | 3 - 207 MB | NPU | [FastSam-X.tflite](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.tflite) |
61
  | FastSam-X | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 32.931 ms | 5 - 86 MB | NPU | [FastSam-X.dlc](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.dlc) |
62
  | FastSam-X | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 35.033 ms | 14 - 98 MB | NPU | [FastSam-X.onnx](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.onnx) |
63
+ | FastSam-X | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | TFLITE | 30.709 ms | 2 - 119 MB | NPU | [FastSam-X.tflite](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.tflite) |
64
  | FastSam-X | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | QNN_DLC | 25.394 ms | 5 - 91 MB | NPU | [FastSam-X.dlc](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.dlc) |
65
  | FastSam-X | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 29.851 ms | 14 - 88 MB | NPU | [FastSam-X.onnx](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.onnx) |
66
  | FastSam-X | float | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 45.562 ms | 355 - 355 MB | NPU | [FastSam-X.dlc](https://huggingface.co/qualcomm/FastSam-X/blob/main/FastSam-X.dlc) |
 
122
  ```bash
123
  python -m qai_hub_models.models.fastsam_x.export
124
  ```
125
+
 
 
 
 
 
 
 
 
 
 
126
 
127
 
128
  ## How does this work?
precompiled/qualcomm-snapdragon-x-elite/FastSam-X.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:cd0651b7874394692e3461d1fbaeff9dbd8810386bb080c93f6a64f855edc8f9
3
  size 149666184
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c03aebd0a72300e9289e81c3b2ee02062e7c88cece5b80f2b26ab0b85b026228
3
  size 149666184
precompiled/qualcomm-snapdragon-x-elite/FastSam-X.onnx.zip CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:aa69ae0dcc71297ee96940c338e50a5179c7b936658abd12630d246290e95780
3
- size 135765548
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:17d32c3698fd6bbda5916de685d6420e20479c975e7a4dd9c820462d8dd753cc
3
+ size 135765543
precompiled/qualcomm-snapdragon-x-elite/sdk_versions.yml ADDED
@@ -0,0 +1,5 @@
 
 
 
 
 
 
1
+ sdk_versions:
2
+ qnn_context_binary:
3
+ qairt: 2.34.2.250528164111_119506
4
+ precompiled_qnn_onnx:
5
+ qairt: 2.33.2.250410134701_117956