Hugging Face's logo

qualcomm
/
Mask2Former

Image Segmentation
PyTorch
android
Model card Files
xet
Community
qaihm-bot commited on
Commit
77e9f39
·
verified ·
1 Parent(s): 3849afb

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +26 -26
README.md CHANGED
@@ -1,6 +1,6 @@
1
  ---
2
  library_name: pytorch
3
- license: apache-2.0
4
  tags:
5
  - android
6
  pipeline_tag: image-segmentation
@@ -25,7 +25,7 @@ More details on model performance across various devices, can be found
25
 
26
  ### Model Details
27
 
28
- - **Model Type:** Semantic segmentation
29
  - **Model Stats:**
30
  - Model checkpoint: facebook/mask2former-swin-tiny-coco-panoptic
31
  - Input resolution: 384x384
@@ -33,24 +33,24 @@ More details on model performance across various devices, can be found
33
  - Model size: 200.6 MB
34
  - Number of output classes: 100
35
 
36
- | Model | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Precision | Primary Compute Unit | Target Model
37
  |---|---|---|---|---|---|---|---|---|
38
- | Mask2Former | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | TFLITE | 1389.603 ms | 162 - 190 MB | FP32 | CPU | [Mask2Former.tflite](https://huggingface.co/qualcomm/Mask2Former/blob/main/Mask2Former.tflite) |
39
- | Mask2Former | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | ONNX | 1044.058 ms | 123 - 127 MB | FP32 | CPU | [Mask2Former.onnx](https://huggingface.co/qualcomm/Mask2Former/blob/main/Mask2Former.onnx) |
40
- | Mask2Former | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | TFLITE | 1086.821 ms | 112 - 144 MB | FP32 | CPU | [Mask2Former.tflite](https://huggingface.co/qualcomm/Mask2Former/blob/main/Mask2Former.tflite) |
41
- | Mask2Former | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | ONNX | 908.724 ms | 207 - 233 MB | FP32 | CPU | [Mask2Former.onnx](https://huggingface.co/qualcomm/Mask2Former/blob/main/Mask2Former.onnx) |
42
- | Mask2Former | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | TFLITE | 882.534 ms | 163 - 183 MB | FP32 | CPU | [Mask2Former.tflite](https://huggingface.co/qualcomm/Mask2Former/blob/main/Mask2Former.tflite) |
43
- | Mask2Former | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | ONNX | 661.962 ms | 177 - 195 MB | FP32 | CPU | [Mask2Former.onnx](https://huggingface.co/qualcomm/Mask2Former/blob/main/Mask2Former.onnx) |
44
- | Mask2Former | SA7255P ADP | SA7255P | TFLITE | 2219.935 ms | 164 - 179 MB | FP32 | CPU | [Mask2Former.tflite](https://huggingface.co/qualcomm/Mask2Former/blob/main/Mask2Former.tflite) |
45
- | Mask2Former | SA8255 (Proxy) | SA8255P Proxy | TFLITE | 1447.171 ms | 140 - 165 MB | FP32 | CPU | [Mask2Former.tflite](https://huggingface.co/qualcomm/Mask2Former/blob/main/Mask2Former.tflite) |
46
- | Mask2Former | SA8295P ADP | SA8295P | TFLITE | 1386.002 ms | 163 - 181 MB | FP32 | CPU | [Mask2Former.tflite](https://huggingface.co/qualcomm/Mask2Former/blob/main/Mask2Former.tflite) |
47
- | Mask2Former | SA8650 (Proxy) | SA8650P Proxy | TFLITE | 1442.301 ms | 138 - 166 MB | FP32 | CPU | [Mask2Former.tflite](https://huggingface.co/qualcomm/Mask2Former/blob/main/Mask2Former.tflite) |
48
- | Mask2Former | SA8775P ADP | SA8775P | TFLITE | 1844.881 ms | 164 - 179 MB | FP32 | CPU | [Mask2Former.tflite](https://huggingface.co/qualcomm/Mask2Former/blob/main/Mask2Former.tflite) |
49
- | Mask2Former | QCS8275 (Proxy) | QCS8275 Proxy | TFLITE | 2219.935 ms | 164 - 179 MB | FP32 | CPU | [Mask2Former.tflite](https://huggingface.co/qualcomm/Mask2Former/blob/main/Mask2Former.tflite) |
50
- | Mask2Former | QCS8550 (Proxy) | QCS8550 Proxy | TFLITE | 1490.226 ms | 164 - 189 MB | FP32 | CPU | [Mask2Former.tflite](https://huggingface.co/qualcomm/Mask2Former/blob/main/Mask2Former.tflite) |
51
- | Mask2Former | QCS9075 (Proxy) | QCS9075 Proxy | TFLITE | 1844.881 ms | 164 - 179 MB | FP32 | CPU | [Mask2Former.tflite](https://huggingface.co/qualcomm/Mask2Former/blob/main/Mask2Former.tflite) |
52
- | Mask2Former | QCS8450 (Proxy) | QCS8450 Proxy | TFLITE | 1636.572 ms | 161 - 186 MB | FP32 | CPU | [Mask2Former.tflite](https://huggingface.co/qualcomm/Mask2Former/blob/main/Mask2Former.tflite) |
53
- | Mask2Former | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 733.104 ms | 267 - 267 MB | FP32 | CPU | [Mask2Former.onnx](https://huggingface.co/qualcomm/Mask2Former/blob/main/Mask2Former.onnx) |
54
 
55
 
56
 
@@ -60,7 +60,7 @@ More details on model performance across various devices, can be found
60
 
61
  Install the package via pip:
62
  ```bash
63
- pip install "qai-hub-models[mask2former]"
64
  ```
65
 
66
 
@@ -112,12 +112,12 @@ python -m qai_hub_models.models.mask2former.export
112
  Profiling Results
113
  ------------------------------------------------------------
114
  Mask2Former
115
- Device : Samsung Galaxy S23 (13)
116
- Runtime : TFLITE
117
- Estimated inference time (ms) : 1389.6
118
- Estimated peak memory usage (MB): [162, 190]
119
- Total # Ops : 3208
120
- Compute Unit(s) : CPU (3208 ops)
121
  ```
122
 
123
 
 
1
  ---
2
  library_name: pytorch
3
+ license: other
4
  tags:
5
  - android
6
  pipeline_tag: image-segmentation
 
25
 
26
  ### Model Details
27
 
28
+ - **Model Type:** Model_use_case.semantic_segmentation
29
  - **Model Stats:**
30
  - Model checkpoint: facebook/mask2former-swin-tiny-coco-panoptic
31
  - Input resolution: 384x384
 
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) | TFLITE | 1905.346 ms | 164 - 178 MB | CPU | [Mask2Former.tflite](https://huggingface.co/qualcomm/Mask2Former/blob/main/Mask2Former.tflite) |
39
+ | Mask2Former | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 2173.199 ms | 159 - 183 MB | CPU | [Mask2Former.tflite](https://huggingface.co/qualcomm/Mask2Former/blob/main/Mask2Former.tflite) |
40
+ | Mask2Former | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 1364.021 ms | 163 - 194 MB | CPU | [Mask2Former.tflite](https://huggingface.co/qualcomm/Mask2Former/blob/main/Mask2Former.tflite) |
41
+ | Mask2Former | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 1595.41 ms | 164 - 178 MB | CPU | [Mask2Former.tflite](https://huggingface.co/qualcomm/Mask2Former/blob/main/Mask2Former.tflite) |
42
+ | Mask2Former | float | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 1905.346 ms | 164 - 178 MB | CPU | [Mask2Former.tflite](https://huggingface.co/qualcomm/Mask2Former/blob/main/Mask2Former.tflite) |
43
+ | Mask2Former | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 1432.0 ms | 123 - 166 MB | CPU | [Mask2Former.tflite](https://huggingface.co/qualcomm/Mask2Former/blob/main/Mask2Former.tflite) |
44
+ | Mask2Former | float | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 1417.48 ms | 157 - 173 MB | CPU | [Mask2Former.tflite](https://huggingface.co/qualcomm/Mask2Former/blob/main/Mask2Former.tflite) |
45
+ | Mask2Former | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 1424.237 ms | 142 - 167 MB | CPU | [Mask2Former.tflite](https://huggingface.co/qualcomm/Mask2Former/blob/main/Mask2Former.tflite) |
46
+ | Mask2Former | float | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 1595.41 ms | 164 - 178 MB | CPU | [Mask2Former.tflite](https://huggingface.co/qualcomm/Mask2Former/blob/main/Mask2Former.tflite) |
47
+ | Mask2Former | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | TFLITE | 1236.454 ms | 157 - 162 MB | CPU | [Mask2Former.tflite](https://huggingface.co/qualcomm/Mask2Former/blob/main/Mask2Former.tflite) |
48
+ | Mask2Former | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | ONNX | 983.166 ms | 113 - 133 MB | CPU | [Mask2Former.onnx](https://huggingface.co/qualcomm/Mask2Former/blob/main/Mask2Former.onnx) |
49
+ | Mask2Former | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 1138.941 ms | 163 - 189 MB | CPU | [Mask2Former.tflite](https://huggingface.co/qualcomm/Mask2Former/blob/main/Mask2Former.tflite) |
50
+ | Mask2Former | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 786.488 ms | 217 - 239 MB | CPU | [Mask2Former.onnx](https://huggingface.co/qualcomm/Mask2Former/blob/main/Mask2Former.onnx) |
51
+ | Mask2Former | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | TFLITE | 878.535 ms | 163 - 177 MB | CPU | [Mask2Former.tflite](https://huggingface.co/qualcomm/Mask2Former/blob/main/Mask2Former.tflite) |
52
+ | Mask2Former | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 694.644 ms | 162 - 176 MB | CPU | [Mask2Former.onnx](https://huggingface.co/qualcomm/Mask2Former/blob/main/Mask2Former.onnx) |
53
+ | Mask2Former | float | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 433.397 ms | 268 - 268 MB | CPU | [Mask2Former.onnx](https://huggingface.co/qualcomm/Mask2Former/blob/main/Mask2Former.onnx) |
54
 
55
 
56
 
 
60
 
61
  Install the package via pip:
62
  ```bash
63
+ pip install "qai-hub-models[mask2former]" git+https://github.com/cocodataset/panopticapi.git
64
  ```
65
 
66
 
 
112
  Profiling Results
113
  ------------------------------------------------------------
114
  Mask2Former
115
+ Device : cs_8275 (ANDROID 14)
116
+ Runtime : TFLITE
117
+ Estimated inference time (ms) : 1905.3
118
+ Estimated peak memory usage (MB): [164, 178]
119
+ Total # Ops : 3213
120
+ Compute Unit(s) : npu (0 ops) gpu (0 ops) cpu (3213 ops)
121
  ```
122
 
123