Hugging Face's logo

qualcomm
/
Mask2Former

Image Segmentation
PyTorch
android
Model card Files
xet
Community
qaihm-bot commited on
Commit
37f53e5
·
verified ·
1 Parent(s): 3063816

v0.32.0

Browse files

See https://github.com/quic/ai-hub-models/releases/v0.32.0 for changelog.

Files changed (4) hide show
  1. .gitattributes +1 -0
  2. DEPLOYMENT_MODEL_LICENSE.pdf +3 -0
  3. LICENSE +2 -0
  4. README.md +20 -22
.gitattributes CHANGED
@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
 
 
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
36
+ DEPLOYMENT_MODEL_LICENSE.pdf filter=lfs diff=lfs merge=lfs -text
DEPLOYMENT_MODEL_LICENSE.pdf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4409f93b0e82531303b3e10f52f1fdfb56467a25f05b7441c6bbd8bb8a64b42c
3
+ size 109629
LICENSE ADDED
@@ -0,0 +1,2 @@
 
 
 
1
+ The license of the original trained model can be found at https://github.com/huggingface/transformers/blob/main/LICENSE.
2
+ The license for the deployable model files (.tflite, .onnx, .dlc, .bin, etc.) can be found in DEPLOYMENT_MODEL_LICENSE.pdf.
README.md CHANGED
@@ -29,29 +29,27 @@ More details on model performance across various devices, can be found
29
  - **Model Stats:**
30
  - Model checkpoint: facebook/mask2former-swin-tiny-coco-panoptic
31
  - Input resolution: 384x384
32
- - Number of parameters: 42M
33
- - Model size: 200.6 MB
34
  - Number of output classes: 100
35
 
36
  | Model | Precision | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit | Target Model
37
  |---|---|---|---|---|---|---|---|---|
38
- | Mask2Former | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_CONTEXT_BINARY | 378.098 ms | 2 - 11 MB | NPU | Use Export Script |
39
- | Mask2Former | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_CONTEXT_BINARY | 352.222 ms | 2 - 18 MB | NPU | Use Export Script |
40
- | Mask2Former | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_CONTEXT_BINARY | 217.892 ms | 2 - 4 MB | NPU | Use Export Script |
41
- | Mask2Former | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_CONTEXT_BINARY | 224.361 ms | 2 - 12 MB | NPU | Use Export Script |
42
- | Mask2Former | float | SA7255P ADP | Qualcomm® SA7255P | QNN_CONTEXT_BINARY | 378.098 ms | 2 - 11 MB | NPU | Use Export Script |
43
- | Mask2Former | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_CONTEXT_BINARY | 217.919 ms | 2 - 4 MB | NPU | Use Export Script |
44
- | Mask2Former | float | SA8295P ADP | Qualcomm® SA8295P | QNN_CONTEXT_BINARY | 288.399 ms | 2 - 17 MB | NPU | Use Export Script |
45
- | Mask2Former | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_CONTEXT_BINARY | 217.23 ms | 2 - 4 MB | NPU | Use Export Script |
46
- | Mask2Former | float | SA8775P ADP | Qualcomm® SA8775P | QNN_CONTEXT_BINARY | 224.361 ms | 2 - 12 MB | NPU | Use Export Script |
47
- | Mask2Former | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | QNN_CONTEXT_BINARY | 217.14 ms | 2 - 5 MB | NPU | Use Export Script |
48
- | Mask2Former | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | PRECOMPILED_QNN_ONNX | 218.642 ms | 0 - 115 MB | NPU | Use Export Script |
49
- | Mask2Former | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_CONTEXT_BINARY | 140.424 ms | 2 - 22 MB | NPU | Use Export Script |
50
- | Mask2Former | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | PRECOMPILED_QNN_ONNX | 140.725 ms | 9 - 29 MB | NPU | Use Export Script |
51
- | Mask2Former | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | QNN_CONTEXT_BINARY | 125.966 ms | 2 - 16 MB | NPU | Use Export Script |
52
- | Mask2Former | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | PRECOMPILED_QNN_ONNX | 106.794 ms | 7 - 21 MB | NPU | Use Export Script |
53
- | Mask2Former | float | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_CONTEXT_BINARY | 224.698 ms | 2 - 2 MB | NPU | Use Export Script |
54
- | Mask2Former | float | Snapdragon X Elite CRD | Snapdragon® X Elite | PRECOMPILED_QNN_ONNX | 224.493 ms | 108 - 108 MB | NPU | Use Export Script |
55
 
56
 
57
 
@@ -116,9 +114,9 @@ Mask2Former
116
  Device : cs_8275 (ANDROID 14)
117
  Runtime : QNN_CONTEXT_BINARY
118
  Estimated inference time (ms) : 378.1
119
- Estimated peak memory usage (MB): [2, 11]
120
- Total # Ops : 1954
121
- Compute Unit(s) : npu (1954 ops) gpu (0 ops) cpu (0 ops)
122
  ```
123
 
124
 
 
29
  - **Model Stats:**
30
  - Model checkpoint: facebook/mask2former-swin-tiny-coco-panoptic
31
  - Input resolution: 384x384
 
 
32
  - Number of output classes: 100
33
 
34
  | Model | Precision | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit | Target Model
35
  |---|---|---|---|---|---|---|---|---|
36
+ | Mask2Former | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_CONTEXT_BINARY | 378.146 ms | 2 - 10 MB | NPU | Use Export Script |
37
+ | Mask2Former | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_CONTEXT_BINARY | 352.578 ms | 2 - 22 MB | NPU | Use Export Script |
38
+ | Mask2Former | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_CONTEXT_BINARY | 218.448 ms | 2 - 5 MB | NPU | Use Export Script |
39
+ | Mask2Former | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_CONTEXT_BINARY | 224.045 ms | 3 - 12 MB | NPU | Use Export Script |
40
+ | Mask2Former | float | SA7255P ADP | Qualcomm® SA7255P | QNN_CONTEXT_BINARY | 378.146 ms | 2 - 10 MB | NPU | Use Export Script |
41
+ | Mask2Former | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_CONTEXT_BINARY | 217.666 ms | 2 - 4 MB | NPU | Use Export Script |
42
+ | Mask2Former | float | SA8295P ADP | Qualcomm® SA8295P | QNN_CONTEXT_BINARY | 286.528 ms | 2 - 18 MB | NPU | Use Export Script |
43
+ | Mask2Former | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_CONTEXT_BINARY | 216.904 ms | 2 - 5 MB | NPU | Use Export Script |
44
+ | Mask2Former | float | SA8775P ADP | Qualcomm® SA8775P | QNN_CONTEXT_BINARY | 224.045 ms | 3 - 12 MB | NPU | Use Export Script |
45
+ | Mask2Former | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | QNN_CONTEXT_BINARY | 216.853 ms | 2 - 4 MB | NPU | Use Export Script |
46
+ | Mask2Former | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | PRECOMPILED_QNN_ONNX | 219.482 ms | 0 - 112 MB | NPU | Use Export Script |
47
+ | Mask2Former | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_CONTEXT_BINARY | 139.643 ms | 2 - 22 MB | NPU | Use Export Script |
48
+ | Mask2Former | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | PRECOMPILED_QNN_ONNX | 140.598 ms | 9 - 23 MB | NPU | Use Export Script |
49
+ | Mask2Former | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | QNN_CONTEXT_BINARY | 125.131 ms | 2 - 16 MB | NPU | Use Export Script |
50
+ | Mask2Former | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | PRECOMPILED_QNN_ONNX | 105.317 ms | 9 - 22 MB | NPU | Use Export Script |
51
+ | Mask2Former | float | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_CONTEXT_BINARY | 224.747 ms | 2 - 2 MB | NPU | Use Export Script |
52
+ | Mask2Former | float | Snapdragon X Elite CRD | Snapdragon® X Elite | PRECOMPILED_QNN_ONNX | 224.607 ms | 108 - 108 MB | NPU | Use Export Script |
53
 
54
 
55
 
 
114
  Device : cs_8275 (ANDROID 14)
115
  Runtime : QNN_CONTEXT_BINARY
116
  Estimated inference time (ms) : 378.1
117
+ Estimated peak memory usage (MB): [2, 10]
118
+ Total # Ops : 1947
119
+ Compute Unit(s) : npu (1947 ops) gpu (0 ops) cpu (0 ops)
120
  ```
121
 
122