Invalid NDArray file format

[MyraBaba]

Hi,

In insightface facemodel.py below code gives error for Arc...model

    if len(args.model)>0:
      self.model = get_model(ctx, image_size, args.model, 'fc1')

mxnet.base.MXNetError: [22:42:18] src/ndarray/ndarray.cc:1834: Check failed: header == kMXAPINDArrayListMagic: Invalid NDArray file format

[tomriddle54]

@1996scarlet and @MyraBaba did you solve this error? Because I am also getting this error

Exception in thread Thread-2:
Traceback (most recent call last):
  File "/media/mustafa/ubuntu_backup/anaconda3/lib/python3.7/threading.py", line 917, in _bootstrap_inner
    self.run()
  File "/media/mustafa/ubuntu_backup/anaconda3/lib/python3.7/threading.py", line 865, in run
    self._target(*self._args, **self._kwargs)
  File "usb_camera.py", line 139, in <lambda>
    Process(target=lambda: asyncio.run(detection_loop(preload))).start()
  File "/media/mustafa/ubuntu_backup/anaconda3/lib/python3.7/asyncio/runners.py", line 43, in run
    return loop.run_until_complete(main)
  File "/media/mustafa/ubuntu_backup/anaconda3/lib/python3.7/asyncio/base_events.py", line 573, in run_until_complete
    return future.result()
  File "usb_camera.py", line 67, in detection_loop
    detector = face_detector.DetectorModel(preload)
  File "/media/mustafa/ubuntu_backup/Projects/mustafa_face_dl/ArcFace-Multiplex-Recognition/face_detector.py", line 14, in __init__
    self.detector = RetinaFace(args.retina_model, 0, args.gpu, 'net3')
  File "/media/mustafa/ubuntu_backup/Projects/mustafa_face_dl/ArcFace-Multiplex-Recognition/retinaface.py", line 84, in __init__
    sym, arg_params, aux_params = mx.model.load_checkpoint(prefix, epoch)
  File "/home/mustafa/.local/lib/python3.7/site-packages/mxnet/model.py", line 438, in load_checkpoint
    save_dict = nd.load('%s-%04d.params' % (prefix, epoch))
  File "/home/mustafa/.local/lib/python3.7/site-packages/mxnet/ndarray/utils.py", line 175, in load
    ctypes.byref(names)))
  File "/home/mustafa/.local/lib/python3.7/site-packages/mxnet/base.py", line 252, in check_call
    raise MXNetError(py_str(_LIB.MXGetLastError()))
mxnet.base.MXNetError: [20:50:08] src/ndarray/ndarray.cc:1803: Check failed: header == kMXAPINDArrayListMagic Invalid NDArray file format

[rebotnix]

@1996scarlet and @MyraBaba
I have the same issue..Do you use mxcu101? Any idea to fix this?

[1996scarlet]

So could you please tell me which version of cuda and mxnet are you using?

Another method worth trying is to use this file to test a single image of the ArcFace model.
https://github.com/deepinsight/insightface/blob/master/deploy/test.py

[tomriddle54]

I have cuda 9.0 and cuda 10.1 but default is 10.1 mxnet-cu101 even tried with mxnet-cu90. I tried the insightface it's working

[yanjieee]

I have tried cuda8.0 and mxnet-cu80, the same error.

[lakshaytalkstomachines]

Download the model again from InsightFace repo. The link is given in the RetinaFace readme.md
I used your model in the RetinaFace code this problem came. But when i redownloaded the model from their repo. the problem vanished. Not sure what is the problem but my naive guess would be corruption of json file.

[jingruhou]

请问你们的MXNet的是哪个版本？

[hoanganh928]

weight files are stored in git lfs, so you must download them to run:
sudo apt-get install git-lfs
git lfs install
git lfs pull

[Raunak-Singh-Inventor]

Hi guys,

I got this error in Linux Mint
mxnet.base.MXNetError: [13:46:01] src/ndarray/ndarray.cc:1112: Check failed: header == kMXAPINDArrayListMagic Invalid NDArray file format
I solved it by doing what @hoanganh928 did above:

sudo apt-get install git-lfs
git lfs install
git lfs pull