Tensorflow model detecting wrong objects?

[msarfrazcss]

Hi, i am new to tensorflow object detection API, i trained a model with 900 images of shoes, i put 20%(180 images) into test forlder and 80%(720 images into train folder), but after training my trained model detecting other objects with shoes, below is the screen shot, anyone can help me please, where i am wrong....

I am training this model on MAC Machine, and using faster_rcnn_inception_v2_coco_2018_01_28 model to train my model, I follow below(Your) link to train my model:

https://github.com/EdjeElectronics/TensorFlow-Object-Detection-API-Tutorial-Train-Multiple-Objects-Windows-10/tree/d1c5b59803543e48362c27c48d704d4b0d92d135

and one more question when i run python on this model on terminal webcam becomes very slow why? (I have no graphics card in my MAC Machine)

Thanks in advance...

[janekzimoch]

I would be also interested in getting to know what might be causing this issue. I had similar problem. I tried to recognise between different types of Gatorade isotonic drinks (6 classes in total), but my model ended up recognising most of bottles as "blue gatorade" with very high confidence, with few cases of "yellow gatorade" (all were incorect) (while the remainign categories were: orange, red, green and bright yellow)

[tanvirahmd1]

A bit irrelevant, how you configured Python path on mac i try to configure python path on mac and its not working, share it with me. For your question may be something is wrong with your data labelling, you can check that

[msarfrazcss]

I set python path by this command:

export PYTHONPATH=/Users/cssjockey-13/tensorflow1/models:/Users/cssjockey-13/tensorflow1/models/research:/Users/cssjockey-13/tensorflow1/models/research/slim

It is working for me.

[tanvirahmd1]

well thanks, i had used many ways but nothing is worked for me on mac, let me try this as well

[lithin-chandran]

i have trained the algorithm in windows and still the same problem is coming up. I am also searching for the reason for this

[BalajiB3663]

Hello these are due to false positive detections, while training your models have you added background images to your training dataset? by doing so you can reduce your false detections.

[msarfrazcss]

Hello these are due to false positive detections, while training your models have you added background images to your training dataset? by doing so you can reduce your false detections.

How can we add background images? Here i have only test and train folder, there is no option to add background images, can you tell me please?

[BalajiB3663]

Refer this link Re: [tensorflow/models] SSD resnet50 fpn model not detecting at all. (#5954)

On Mon 7 Jan, 2019, 10:10 msarfrazcss <notifications@github.com wrote:

Hello these are due to false positive detections, while training your models have you added background images to your training dataset? by doing so you can reduce your false detections.

How can we add background images? Here i have only test and train folder, there is no option to add background images, can you tell me please?

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/EdjeElectronics/TensorFlow-Object-Detection-API-Tutorial-Train-Multiple-Objects-Windows-10/issues/175#issuecomment-451820227, or mute the thread https://github.com/notifications/unsubscribe-auth/Abt0-NxOFPzGk50WUdz_vQjZOYC8JL3dks5vAs_VgaJpZM4X3LVa .

[mehmetsoylu]

Hello @msarfrazcss , i have the same problem with yours. Did you solved your problem? If you solved it, how did you solved it? Can you please explain to me?

[ihtisham1211]

i also need the solution

[ttducqt]

i got the same problem

[amyadamslou]

I'm facing the same issue too.. Did anyone figure this out?