Error executing [ap, recall, precision] = evaluateDe​tectionPre​cision(res​ults, expectedResults);

I have modified the FasterRCNN tutorial and have successfully created a detector, however when I get to the testing step I cannot execute the final code. The errors I get are below :
Error using vision.internal.detector.evaluationInputValidation>checkGroundTruthTable (line 101)
Invalid bounding box in row 1, column 1 of the training data table: Cell contents reference from a non-cell array
object.
Error in vision.internal.detector.evaluationInputValidation (line 8)
checkGroundTruthTable(groundTruth, height(detectionResults), mfilename);
Error in evaluateDetectionPrecision (line 94)
vision.internal.detector.evaluationInputValidation(detectionResults, ...
Error in RAT_DeepLearningFasterRCNNObjectDetectionExample (line 318)
[ap, recall, precision] = evaluateDetectionPrecision(results, expectedResults);
I have looked my results table and it seems to be in the correct format, with the boxes, the scores for each and the labels. What could be causing this error?

답변 (1개)

The second parameter of the 'evaluateDetectionPrecision' function should be in the format as specified in the documentation link below:
The most likely cause of the error message: "Cell contents reference from a non-cell array" is when it is referenced incorrectly as below:
>> d = [ 1 2 4 5];
>> d{1}
Cell contents reference from a non-cell array object.
Ensure that the second parameter is in the right format and is referenced correctly.

댓글 수: 1

I think my variable is incorrectly formatted but I cannot figure out how to convert it. I think it is cells instead of a double but the cell2mat function does not seem to be working. Does anyone know how to fix my table or will I have to redraw all of my ROIs? I did attempt to reimport the ROIs I have already drawn and reexport them but they stayed in the incorrect format. But when I start a new drawing session and export those, they export in the correct format.
I have attached a screenshot of the structure of my table.

댓글을 달려면 로그인하십시오.

카테고리

도움말 센터File Exchange에서 Numeric Types에 대해 자세히 알아보기

질문:

2017년 4월 26일

편집:

2017년 5월 15일

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by