how to get the output value including the tuple?
khseob0715 opened this issue · comments
Hanseob Kim commented
I have used this library before.
Previously, it was easy to use because only one result was obtained.
I am currently trying to utilize Mask RCNN in unity
I've written some code, but there's a problem with the way I get the output.
`
runner.AddInput(graph["input_image"][0], input_image_tensor)
.AddInput(graph["input_image_meta"][0], input_image_meta_tensor)
.Fetch(graph["mrcnn_detection/Reshape_1"][0])
.Fetch(graph["mrcnn_class/Reshape_1"][0])
.Fetch(graph["mrcnn_bbox/Reshape"][0])
.Fetch(graph["mrcnn_mask/Reshape_1"][0]);
TFTensor mrcnn_detection = output[0];
// Returns: [batch, num_detections, (y1, x1, y2, x2, class_id, score)] in python model.py
TFTensor mrcnn_class = output[1];
TFTensor mrcnn_bbok = output[2];
TFTensor mrcnn_mask = output[3];
// Returns: Masks [batch, num_rois, MASK_POOL_SIZE, MASK_POOL_SIZE, NUM_CLASSES] in python model.py
Debug.Log(mrcnn_detection.GetValue(jagged: true).GetType()); // System.Single[][][]
`
I want to get the return value of detection, which is the first output.
This returns (float float tuple) at model.py on the mask rcnnn.
But, the Getvalue() returned the single[][][] types.
I don't know how to handle this output.
Please give me some advice.
Thank you in advance.
Chrigui Mohamed commented
Hello @khseob0715,
did you find any solution to handle MaskRcnn model using TensorflowSharp? I have the same case and I am blocked.