DrewNF / Tensorflow_Object_Tracking_Video

Object Tracking in Tensorflow ( Localization Detection Classification ) developed to partecipate to ImageNET VID competition
MIT License
503 stars 198 forks source link

AttributeError: 'module' object has no attribute 'add_rectangles #1

Open violivei opened 8 years ago

violivei commented 8 years ago

Hello,

I'm testing your code for videos and when I executed the script "train_multiclass.py" to obtain the model, it gave me the following error:

Traceback (most recent call last):
  File "TENSORBOX/train_multiclass.py", line 610, in <module>
    main()
  File "TENSORBOX/train_multiclass.py", line 607, in main
    train(H, test_images=[])
  File "TENSORBOX/train_multiclass.py", line 555, in train
    test_output_to_log = train_utils_multiclass.add_rectangles(H,
AttributeError: 'module' object has no attribute 'add_rectangles'

Is this expected?

Probably we should use "train_utils.add_rectangles" I guess?

Thanks in advance.

DrewNF commented 8 years ago

I update the project, knwo there is only one train file and you set the number of class in the hypes file, there's no difference between train on a single or multi class. Hope this solve your problem.

violivei commented 8 years ago

Hello @DrewNF

I saw your updates. Thanks a lot. However, I'm having real issues trying to run this with multiclass images.

Is this a good example?

"omo_section6_27.png" (32.0,196.0,138.0,258.0) :1 /1,(136.0,193.0,243.0,255.0) :1 /1,(34.0,255.0,134.0,325.0) :1 /1,(142.0,253.0,245.0,329.0) :1 /1,(247.0,256.0,348.0,324.0) :1 /1,(9.0,352.0,112.0,411.0) :1 /1,(14.0,417.0,114.0,475.0) :1 /1,(116.0,350.0,217.0,415.0) :1 /1,(114.0,412.0,214.0,481.0) :1 /1,(218.0,414.0,315.0,479.0) :1 /1,(322.0,354.0,410.0,409.0) :1 /1,(320.0,409.0,409.0,475.0) :1 /1,(421.0,217.0,493.0,279.0) :1 /1,(422.0,284.0,495.0,353.0) :1 /1,(501.0,280.0,609.0,348.0) :1 /1,(414.0,383.0,476.0,451.0) :1 /1;
"911468_1273803_ORAL CARE_20160707_CategoryView.jpg_24.png" (24,38,122,58) :1 /2,(29,60,122,78) :1 /2,(24,81,125,103) :1 /2,(22,106,124,117) :1 /2,(125,37,210,57) :1 /2,(122,61,212,79) :1 /2,(127,82,212,102) :1 /2,(125,103,214,118) :1 /2,(244,34,349,54) :1 /2,(236,56,342,82) :1 /2,(239,82,343,108) :1 /2,(243,104,342,120) :1 /2,(349,37,372,55) :1 /2,(347,56,371,75) :1 /2,(344,77,369,93) :1 /2,(345,95,370,113) :1 /2,(347,113,370,130) :1 /2,(370,18,455,42) :1 /2,(373,41,454,63) :1 /2,(374,62,455,83) :1 /2,(457,26,536,43) :1 

I'm expecting black and white boxes printed in the test images but only the black ones appear. I understand that silhouetteid is actually the class label, but what about the confidencescore?

DrewNF commented 8 years ago

For know the multiclass recognize and draw rectangles on different objects but really don't recognize wich ones are. I'm working to adapt inception to get the real label of the object. This is an issue that i'm trying to solve. But I will realise it after the submission deadline of ImageNET :)