Test with small model (4 floats input): works perfectly
Test with MobileFaceNet (37632 floats input):
- inputData(0) = -0.945 (correct)
- After CopyFromBuffer, reading back: 1.32E18 (corrupted!)
Is there a size limit for TfLiteTensorCopyFromBuffer ?
Is there another way to fill large input tensors ?
Thx
Test with MobileFaceNet (37632 floats input):
- inputData(0) = -0.945 (correct)
- After CopyFromBuffer, reading back: 1.32E18 (corrupted!)
Is there a size limit for TfLiteTensorCopyFromBuffer ?
Is there another way to fill large input tensors ?
Thx