Open mauricelos opened 4 years ago
@richardaecn kindly share following examples :
It will be very helpful for us to see as how we prepare data and train Attribute Mask RCNN network.
Thank You
@mitraavi @mauricelos I was Not able to run the model using inference.py https://github.com/tensorflow/tpu/tree/master/models/official/detection/projects/fashionpedia I cant import such as from utils import box_utils from utils import input_utils from utils import mask_utils from utils.object_detection import visualization_utils How do you inference?
@22wei22 and @SikandAlex Your problems with inference have nothing to do with the issue I'm reporting here. Please open a separate issue for yours.
@mauricelos do you find out how to work around it? Also could you let me know how you run the inference code? I can see that this is not related to your post but the authors of the code are not responding at all so if you could support here?
Hi, I'm currently trying to wrap my head around how to initiate training with the fashionpedia dataset. I was able to run the model using inference.py and pre trained weights, but training is a mystery to me. I looked at this GETTING STARTED and for training we need to provide tfrecord files, but the Fashionpedia dataset is using the Coco dataset format. How do I training a model on this dataset or how can I convert the fashionpedia dataset to tfrecords files?
Edit: I tried converting the fashionpedia dataset to tfrecords using a modified version of create_coco_tf_record.py, unfortunately this didn't work. I get this error at the moment:
Which I suspect to be an issue with my tfrecord files, it's clearly expecting the shape of length of all the fashionpedia attributes, where as I probably only have the attributes found in one of the images. The structure of the input tfrecords remains unclear to me.
Thanks in advance!