Updated reference for mobilenet-yolo-v4-syg model (#3222)

Author: Anna Grebneva

models/public/index.md

@@ -115,7 +115,7 @@ SSD-based and provide reasonable accuracy/performance trade-offs.
 | Faster R-CNN with Inception-ResNet v2| TensorFlow\*             | [faster_rcnn_inception_resnet_v2_atrous_coco](./faster_rcnn_inception_resnet_v2_atrous_coco/README.md)| 40.69% | 30.687 | 13.307 |
 | Faster R-CNN with ResNet 50          | TensorFlow\*             | [faster_rcnn_resnet50_coco](./faster_rcnn_resnet50_coco/README.md) | 31.09% | 57.203 | 29.162 |
 | MobileFace Detection V1              | MXNet\*                  | [mobilefacedet-v1-mxnet](./mobilefacedet-v1-mxnet/README.md)| 	78.7488%| 3.5456 | 7.6828 |
-| Mobilenet-yolo-v4-syg                | Keras\*                  | [mobilenet-yolo-v4-syg](./mobilenet-yolo-v4-syg/README.md)| 	84.44%| 65.981 | 61.922 |
+| Mobilenet-yolo-v4-syg                | Keras\*                  | [mobilenet-yolo-v4-syg](./mobilenet-yolo-v4-syg/README.md)| 	86.35%| 65.981 | 61.922 |
 | MTCNN                                | Caffe\*                  | [mtcnn](./mtcnn/README.md):<br>mtcnn-p <br>mtcnn-r <br>mtcnn-o| 48.1308%/62.2625% | <br>3.3715<br>0.0031<br>0.0263|<br>0.0066<br>0.1002<br>0.3890|
 | Pelee                                | Caffe\*                  | [pelee-coco](./pelee-coco/README.md) | 21.9761% | 1.290 | 5.98 |
 | RetinaFace with ResNet 50            | PyTorch\*                | [retinaface-resnet50-pytorch](./retinaface-resnet50-pytorch/README.md) | 91.78% | 88.8627 | 27.2646 |

models/public/mobilenet-yolo-v4-syg/README.md

@@ -23,7 +23,7 @@ which is our own made\* validation dataset for converted model.
 
 | Metric |  Value |
 | ------ | -------|
-| mAP    | 87.11% |
+| mAP    | 86.35% |
 
 ## Input
 

models/public/mobilenet-yolo-v4-syg/accuracy-check.yml

@@ -87,4 +87,4 @@ models:
             integral: 11point
             ignore_difficult: true
             presenter:  print_scalar
-            reference: 0.8711
+            reference: 0.8635