v0.32.0

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

.gitattributes

@@ -34,3 +34,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
 EfficientViT-l2-seg.so filter=lfs diff=lfs merge=lfs -text
+DEPLOYMENT_MODEL_LICENSE.pdf filter=lfs diff=lfs merge=lfs -text

DEPLOYMENT_MODEL_LICENSE.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4409f93b0e82531303b3e10f52f1fdfb56467a25f05b7441c6bbd8bb8a64b42c
+size 109629

EfficientViT-l2-seg.onnx

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:93a9070b1f96082aff260e4573a0ca02b2dac31fa90b85e145efed1596c53fae
-size 210499418
+oid sha256:a94ca3a165666c4bf07b4b4db6f0d0cf1a45637226f57a2f6c477a21122b2fd8
+size 210513688

LICENSE

@@ -0,0 +1,2 @@
+The license of the original trained model can be found at https://github.com/CVHub520/efficientvit/blob/main/LICENSE.
+The license for the deployable model files (.tflite, .onnx, .dlc, .bin, etc.) can be found in DEPLOYMENT_MODEL_LICENSE.pdf.

README.md

@@ -30,16 +30,15 @@ More details on model performance across various devices, can be found
 - **Model Stats:**
   - Model checkpoint: l2.pt
   - Input resolution: 2048x1024
-  - Number of parameters: 53M
   - Number of output classes: 19
-  - Model size (float): 200 MB
+  - Model size (float): 201 MB
 
 | Model | Precision | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit | Target Model
 |---|---|---|---|---|---|---|---|---|
-| EfficientViT-l2-seg | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | ONNX | 1900.82 ms | 6 - 304 MB | NPU | [EfficientViT-l2-seg.onnx](https://huggingface.co/qualcomm/EfficientViT-l2-seg/blob/main/EfficientViT-l2-seg.onnx) |
-| EfficientViT-l2-seg | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 1729.898 ms | 126 - 1004 MB | NPU | [EfficientViT-l2-seg.onnx](https://huggingface.co/qualcomm/EfficientViT-l2-seg/blob/main/EfficientViT-l2-seg.onnx) |
-| EfficientViT-l2-seg | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 1230.575 ms | 103 - 1058 MB | NPU | [EfficientViT-l2-seg.onnx](https://huggingface.co/qualcomm/EfficientViT-l2-seg/blob/main/EfficientViT-l2-seg.onnx) |
-| EfficientViT-l2-seg | float | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 2416.342 ms | 144 - 144 MB | NPU | [EfficientViT-l2-seg.onnx](https://huggingface.co/qualcomm/EfficientViT-l2-seg/blob/main/EfficientViT-l2-seg.onnx) |
+| EfficientViT-l2-seg | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | ONNX | 2089.596 ms | 4 - 269 MB | NPU | [EfficientViT-l2-seg.onnx](https://huggingface.co/qualcomm/EfficientViT-l2-seg/blob/main/EfficientViT-l2-seg.onnx) |
+| EfficientViT-l2-seg | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 2163.163 ms | 131 - 1007 MB | NPU | [EfficientViT-l2-seg.onnx](https://huggingface.co/qualcomm/EfficientViT-l2-seg/blob/main/EfficientViT-l2-seg.onnx) |
+| EfficientViT-l2-seg | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 1231.661 ms | 104 - 1059 MB | NPU | [EfficientViT-l2-seg.onnx](https://huggingface.co/qualcomm/EfficientViT-l2-seg/blob/main/EfficientViT-l2-seg.onnx) |
+| EfficientViT-l2-seg | float | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 2388.516 ms | 144 - 144 MB | NPU | [EfficientViT-l2-seg.onnx](https://huggingface.co/qualcomm/EfficientViT-l2-seg/blob/main/EfficientViT-l2-seg.onnx) |
 
 
 
@@ -103,8 +102,8 @@ Profiling Results
 EfficientViT-l2-seg
 Device                          : cs_8_gen_2 (ANDROID 13)              
 Runtime                         : ONNX                                 
-Estimated inference time (ms)   : 1900.8                               
-Estimated peak memory usage (MB): [6, 304]                             
+Estimated inference time (ms)   : 2089.6                               
+Estimated peak memory usage (MB): [4, 269]                             
 Total # Ops                     : 464                                  
 Compute Unit(s)                 : npu (462 ops) gpu (0 ops) cpu (2 ops)
 ```