[tensor_trainer] Fixed output tensor's dimension and type
authorhyunil park <hyunil46.park@samsung.com>
Mon, 26 Dec 2022 06:22:31 +0000 (15:22 +0900)
committerMyungJoo Ham <myungjoo.ham@samsung.com>
Thu, 2 Feb 2023 08:30:19 +0000 (17:30 +0900)
commitf33d63e8ef3c76167642d07c538d350e33368275
tree790521bd3992b1212b7eb97d324b0b5741a52a53
parentf2c03813114f99ed90cb6f3fc68abdff6ea4e761
[tensor_trainer] Fixed output tensor's dimension and type

Remove the properties that determines the output tensor's caps

- Remove output and outputtype property
- Add initialization of output tensor's dimension and type
- output dimension is 1:1:4:1, '4' has the value of loss, accuracy,
  val_loss, val_accuracy
- output type is float16
- Modify example launch line and log

Signed-off-by: hyunil park <hyunil46.park@samsung.com>
gst/nnstreamer/elements/gsttensor_trainer.c
gst/nnstreamer/elements/gsttensor_trainer.h