Readme is updated/ minor refactoring

Author: daniil-lyakhov

examples/models/yolo12/README.md

@@ -13,13 +13,15 @@ This example demonstrates how to perform inference of [Ultralytics YOLO12 family
 # Performance Evaluation
 
 | CPU                            | Model   | Backend  | Device | Precision | Average Latency, ms |
-|--------------------------------|---------|----------|--------|-----------|-----------------|
-| Intel(R) Core(TM) Ultra 7 155H | yolo12s | openvino | CPU    | FP32      | 88.3549         |
-| Intel(R) Core(TM) Ultra 7 155H | yolo12s | openvino | CPU    | INT8      | 53.066          |
-| Intel(R) Core(TM) Ultra 7 155H | yolo12l | openvino | CPU    | FP32      | 317.953         |
-| Intel(R) Core(TM) Ultra 7 155H | yolo12s | openvino | CPU    | INT8      | 150.846         |
-| Intel(R) Core(TM) Ultra 7 155H | yolo12s | xnnpack  | CPU    | FP32      | 169.36          |
-| Intel(R) Core(TM) Ultra 7 155H | yolo12l | xnnpack  | CPU    | FP32      | 436.876         |
+|--------------------------------|---------|----------|--------|-----------|---------------------|
+| Intel(R) Core(TM) Ultra 7 155H | yolo12s | openvino | CPU    | FP32      | 88.3549             |
+| Intel(R) Core(TM) Ultra 7 155H | yolo12s | openvino | CPU    | INT8      | 53.066              |
+| Intel(R) Core(TM) Ultra 7 155H | yolo12l | openvino | CPU    | FP32      | 317.953             |
+| Intel(R) Core(TM) Ultra 7 155H | yolo12s | openvino | CPU    | INT8      | 150.846             |
+| Intel(R) Core(TM) Ultra 7 155H | yolo12s | openvino | GPU    | FP32      | 32.71               |
+| Intel(R) Core(TM) Ultra 7 155H | yolo12l | openvino | GPU    | FP32      | 70.885              |
+| Intel(R) Core(TM) Ultra 7 155H | yolo12s | xnnpack  | CPU    | FP32      | 169.36              |
+| Intel(R) Core(TM) Ultra 7 155H | yolo12l | xnnpack  | CPU    | FP32      | 436.876             |
 
 
 # Instructions
@@ -63,6 +65,7 @@ XNNPACK:
 python export_and_validate.py --model_name yolo12s --input_dims=[1920,1080] --backend xnnpack
 ```
 
+> **_NOTE:_**  Quantization for XNNPACK backend is WIP. Please refere to https://github.com/pytorch/executorch/issues/11523 for more details.
 
 Exported model could be validated using the `--validate` key:
 

examples/models/yolo12/export_and_validate.py

@@ -68,11 +68,6 @@ def __len__(self):
         return len(self._iter)
 
 
-def visualize_fx_model(model: torch.fx.GraphModule, output_svg_path: str):
-    g = FxGraphDrawer(model, output_svg_path)
-    g.get_dot_graph().write_svg(output_svg_path)
-
-
 def lower_to_openvino(
     aten_dialect: ExportedProgram,
     example_args: Tuple[Any, ...],
@@ -110,7 +105,6 @@ def ext_transform_fn(sample):
                 fold_quantize=False,
             )
 
-            visualize_fx_model(quantized_model, "tmp_quantized_model.svg")
             aten_dialect = torch.export.export(quantized_model, example_args)
             # Convert to edge dialect and lower the module to the backend with a custom partitioner
         compile_spec = [CompileSpec("device", device.encode())]