Fatal: Unsupported image channels: 192

[maziyi234]

(tensorflow) F:\code\nncase\examples\20classes_yolo>"F:\code\nncase\examples\20classes_yolo\ncc_win_x86_64\ncc.exe" compile model/yolov3.tflite k210/kpu_20classes_example/yolo.kmodel -i tflite -o kmodel -t k210 --dataset images2

1. Import graph...
2. Optimize Pass 1...
3. Optimize Pass 2...
4. Quantize... 4.1. Add quantization checkpoints... 4.2. Get activation ranges... Plan buffers... Run calibration... Fatal: Unsupported image channels: 192 ] 0% 0s

你好，我在yolov3中tfliite转K210模型得时候，出现了以上问题，我测试例程得时候没问题，但是用自己训练出来的模型，有问题，tflite模型可以测试，另外，我如何将yolov3能加载到MAIX DOCK板卡上呢

[curioyang]

输出数据layout错了吧

[maziyi234]

"F:\code\nncase\examples\20classes_yolo\ncc_win_x86_64\ncc.exe" compile model/yolov3.tflite k210/kpu_20classes_example/yolo.kmodel -i tflite -o kmodel -t k210 --dataset images2

这个我应该怎么改呢

[curioyang]

"F:\code\nncase\examples\20classes_yolo\ncc_win_x86_64\ncc.exe" compile model/yolov3.tflite k210/kpu_20classes_example/yolo.kmodel -i tflite -o kmodel -t k210 --dataset images2

这个我应该怎么改呢

用最新的nncase python的方式吧， ncc很久不更新了

[curioyang]

@maziyi234 https://github.com/kendryte/nncase/blob/master/docs/USAGE_EN.md#compile-uint8-model-for-tflite If any other problems, reopen the issue.