Enable TinyYolo v3 in CI (#3651)

ngraph/python/tests/__init__.py

@@ -208,7 +208,6 @@ xfail_issue_39661 = xfail_test(reason="RuntimeError: NonMaxSuppression operation
 xfail_issue_39662 = xfail_test(reason="RuntimeError: 'ScatterElementsUpdate' layer with name 'y' have "
                                       "indices value that points to non-existing output tensor element")
 xfail_issue_39663 = xfail_test(reason="RuntimeError: Unsupported primitive of type: ROIAlign name: Y")
-xfail_issue_43380 = xfail_test(reason="RuntimeError: Sorting not possible, due to existed loop")
 xfail_issue_41894 = xfail_test(reason="CPU plugin elementwise computation missmatch")
 
 

ngraph/python/tests/test_onnx/test_zoo_models.py

@@ -28,7 +28,6 @@ from tests.test_onnx.utils.model_importer import ModelImportRunner
 from tests import (
     xfail_issue_38701,
     xfail_issue_43742,
-    xfail_issue_43380,
     xfail_issue_45457,
     xfail_issue_40957,
     xfail_issue_37957,
@@ -52,8 +51,18 @@ def yolov3_post_processing(outputs : Sequence[Any]) -> Sequence[Any]:
     outputs[concat_out_index] = concat_out
     return outputs
 
+def tinyyolov3_post_processing(outputs : Sequence[Any]) -> Sequence[Any]:
+    concat_out_index = 2
+    # remove all elements with value -1 from yolonms_layer_1:1 output
+    concat_out = outputs[concat_out_index][outputs[concat_out_index] != -1]
+    concat_out = concat_out.reshape((outputs[concat_out_index].shape[0], -1, 3))
+    outputs[concat_out_index] = concat_out
+    return outputs
+
 post_processing = {
-    "yolov3" : {"post_processing" : yolov3_post_processing}
+    "yolov3" : {"post_processing" : yolov3_post_processing},
+    "tinyyolov3" : {"post_processing" : tinyyolov3_post_processing},
+    "tiny-yolov3-11": {"post_processing": tinyyolov3_post_processing},
 }
 
 tolerance_map = {
@@ -114,6 +123,8 @@ tolerance_map = {
     "test_mobilenetv2-1": {"atol": 1e-04, "rtol": 0.001},
     "yolov3": {"atol": 0.001, "rtol": 0.001},
     "yolov4": {"atol": 1e-04, "rtol": 0.001},
+    "tinyyolov3": {"atol": 1e-04, "rtol": 0.001},
+    "tiny-yolov3-11": {"atol": 1e-04, "rtol": 0.001},
 }
 
 zoo_models = []
@@ -173,8 +184,6 @@ if len(zoo_models) > 0:
             (xfail_issue_39669, "test_onnx_model_zoo_text_machine_comprehension_t5_model_t5_encoder_12_t5_encoder_cpu"),
             (xfail_issue_38084, "test_onnx_model_zoo_vision_object_detection_segmentation_mask_rcnn_model_MaskRCNN_10_mask_rcnn_R_50_FPN_1x_cpu"),
             (xfail_issue_38084, "test_onnx_model_zoo_vision_object_detection_segmentation_faster_rcnn_model_FasterRCNN_10_faster_rcnn_R_50_FPN_1x_cpu"),
-            (xfail_issue_43380, "test_onnx_model_zoo_vision_object_detection_segmentation_tiny_yolov3_model_tiny_yolov3_11_yolov3_tiny_cpu"),
-            (xfail_issue_45457, "test_MSFT_opset10_mlperf_ssd_resnet34_1200_ssd_resnet34_mAP_20.2_cpu"),
 
             # Model MSFT
             (xfail_issue_37973, "test_MSFT_opset7_tf_inception_v2_model_cpu"),
@@ -191,8 +200,7 @@ if len(zoo_models) > 0:
 
             (xfail_issue_39669, "test_MSFT_opset9_cgan_cgan_cpu"),
             (xfail_issue_40957, "test_MSFT_opset10_BERT_Squad_bertsquad10_cpu"),
-
-            (xfail_issue_43380, "test_MSFT_opset11_tinyyolov3_yolov3_tiny_cpu")
+            (xfail_issue_45457, "test_MSFT_opset10_mlperf_ssd_resnet34_1200_ssd_resnet34_mAP_20.2_cpu"),
 
         ]
         for test_case in import_xfail_list + execution_xfail_list: