our repeat step for training and testing(over)

[gmt710]

our repeat step for training and testing

windows(only cpu) python3.6 keras==2.2.4 keras-resnet six scipy Pillow pandas tensorflow==1.12.0 tqdm opencv-python==3.2.0.6

or

ubuntu16.04
python3.5 keras==2.2.4 keras-resnet six scipy Pillow pandas tensorflow-gpu tqdm opencv-python==3.1.0.4

our repeat step for training

• [x] 1.copy ./object_detector_retinanet/keras_retinanet_bin/train.py class_mappings.csv to ./

• [x] 2.modify ./object_detector_retinanet/utils.py line42 or 45 to our own datapath

• [x] 3.add 2 lines in (./object_detector_retinanet/utils/image.py line24-25 issues23

  from PIL import ImageFile                         
  ImageFile.LOAD_TRUNCATED_IMAGES = True                                     

• [x] 4.in our datapath/annotations/ delete the first line of csv file

• [x] 5.train 1st step run at ./

python -u ./train.py csv

Display for windows:

It's a slow response to downloading the h5 model.

Display for linux:

Training:

• [ ] 6.train 2nd step run at ./

  cp ./object_detector_retinanet/keras_retinanet/bin/train_iou.py ./
  python -u ./train_iou.py --weights snapshot/Tue_May_26_08_17_14_2020/resnet50_csv_05.h5 csv

  ​

• [ ] 7 copy ./object_detector_retinanet/keras_retinanet_bin/predict.py to ./

• [ ] 8.test run at ./ may we'll met this(“ 'dict' object has no attribute 'iteritems' ”):

  vim ./object_detector_retinanet/keras_retinanet/utils/CollapsingMoG.py
  line 190
  change:
  for j, t_vals in clusters.iteritems():
  to:
  for j, t_vals in clusters.items():

  test

  python -u ./predict.py csv  "./snapshot/Thu_May_28_02:36:24_2020/iou_resnet50_csv_01.h5" --hard_score_rate=0.5 

  we'll get detection result.

First three step I've changed.The second step you need modify yourself.

[eg4000]

Thanks for sharing. Did it eventually work?

[gmt710]

Further validation may be required as I am downloading h5 file. I will update the issue whether it succeeds or not.

[gmt710]

Ok.I am training now.

[xuanwdx]

Here, I met a programme trouble as below.

I searched all the code carefully but couldn't find _strided_slice1 out. Would you mind having a look at this? Thank you.

[eg4000]

Hi, I think it's a version issue in RetinaNet. Did you search for issues in their repository?

https://github.com/fizyr/keras-retinanet/issues/1118

Also, here: https://github.com/eg4000/SKU110K_CVPR19/issues/51