AttributeError: 'NoneType' object has no attribute 'TF_NewStatus' about ggcnn HOT 6 CLOSED

Hi,
So it looks like there's two different issues. The first is the missing hdf5 file (which I think you've resolved by creating your own and changing the path, is that correct?).
The second one looks like it's potentially an issue with your tensorflow/keras install, which I'm not so sure about, but seems to come up in a quick google, e.g. tensorflow/tensorflow#8652

from ggcnn.

I didn't train with the train.py
Firstly, I run the download_pretrained_ggcnn.sh to generate the dataset_190502_2135.hdf5 in dataset file,

Secondly ,i run the evluate.py, then occur the error:
IOError: Unable to open file (unable to open file: name = '/home/douglas/dev/ae_grasp_prediction/data/datasets/dataset_rotated_width_zoom_171219_1516.hdf5', errno = 2, error message = 'No such file or directory', flags = 0, o_flags = 0)
So i modify the path and i dont't find the dataset_rotated_width_zoom_171219_1516.hdf5, so i replace it with the dataset_190502_2135.hdf5 in the _dataset.txt.

Finally, i run the evaluate.py, the error occur is that:
Traceback (most recent call last):
File "evaluate.py", line 196, in
run()
File "evaluate.py", line 173, in run
succeeded, failed = calculate_iou_matches(grasp_positions_out, grasp_angles_out, bbs_all, no_grasps=NO_GRASPS, grasp_width_out=grasp_width_out)
File "evaluate.py", line 102, in calculate_iou_matches
gt_bbs = BoundingBoxes.load_from_array(ground_truth_bbs[i, ].squeeze())
IndexError: index 0 is out of bounds for axis 0 with size 0

This is my step, i didn't run the train.py, can i run the evaluate.py directly? i can't get the result...
And my environment is :
tensorflow1.10
keras:2.0.6
Thank you very much!

from ggcnn.

Hi, I try run the train_ggcnn.py,the error is :
ValueError: Input 0 is incompatible with layer conv2d_1: expected ndim=4, found ndim=2
Could you please tell me some details about it?I am looing forward to your reply.

from ggcnn.

Hi,
So it looks like there's two different issues. The first is the missing hdf5 file (which I think you've resolved by creating your own and changing the path, is that correct?).
The second one looks like it's potentially an issue with your tensorflow/keras install, which I'm not so sure about, but seems to come up in a quick google, e.g. tensorflow/tensorflow#8652

Thank you very much for your help.And I solve this problem recently. You are right, and the path of my cornell dataset is wrong.And I run the evaluate.py , then i get the reult such as :
190503_1859__ggcnn_9_5_3__32_16_8 5 1051.0/1760.0 59.72%
How can i input the image to get the result of your paper (bounding box)

from ggcnn.

I download the data from http://s.dougsm.com/ggcnn_rss/ just like' download_pretrained_ggcnn.sh' do.After I unzip it I can't find 'dataset_rotated_width_zoom_171219_1516.hdf5' or 'dataset_190502_1033.hdf5'. Is there something I done wrong?

from ggcnn.

Hello!
If you're still having trouble, have a look at the new version that I just pushed. It will require a little bit of set-up (it uses PyTorch rather than Tensorflow), but should be a lot more user friendly. There might still be some issues to work out, so let me know if there is problems.

from ggcnn.