Trending Tutorials

How To Train an Object Detection Classifier Using TensorFlow 1.5 (GPU) on Windows 10



These instructions work for newer versions of TensorFlow too! This tutorial shows you how to train your own object detector for multiple objects using Google’s TensorFlow Object Detection API on Windows. A written version of the tutorial is available at .

Don’t have time to go through this process, or don’t have a computer powerful enough to complete training quickly? Let me train…

22 Comments on How To Train an Object Detection Classifier Using TensorFlow 1.5 (GPU) on Windows 10

  1. Hey everyone! I recently updated the written version of this guide to work with TensorFlow versions up to 1.13.1. If you are encountering errors following this video, please check out the guide and make sure you are using the most up-to-date commands. Here are answers to some common questions:
    – Where is the train.py file? The train.py file is now located in the /object_detection/legacy folder. You can copy it to the /object_detection folder and use it as normal.
    – Images aren't appearing when testing the Jupyter notebook, even though there were no errors? Try this: go in to object_detection/utils/visualization_utils.py and comment out the import statements on line 25 and 26 that include matplotlib. Then, re-run the script.
    – Can I run this just using CPU and not GPU? (Yes, just use "pip install tensorflow" instead of "pip install tensorflow-gpu".)
    – How to resolve errors related to _pb2 files? See Step 2f of the guide linked below.
    Check out the guide at https://github.com/EdjeElectronics/TensorFlow-Object-Detection-API-Tutorial-Train-Multiple-Objects-Windows-10 !

  2. Is set PATH is necessary to do every time when we close the previous conda?

  3. hey guys if i use " SSDLite Mobilenet v2 " with this tutorial what would be the Exporting the inference graph command? please help

  4. I have amd graphic card can I be able to run tensorflow or I need an Nvidia graphic card for it

  5. Wat graphics card you used in your project. Which graphics card suits me at for this project at lower price point thank you

  6. is it possible to get coordinates out of each object?

  7. Everything works fine but my video frame is reading very slow there is a delay of 2 seconds in reading every frame.
    i am running this code on lenevo Y700, is it because my system configuration is low?

  8. hey thanks for @Edje Electronics for posting great tutorial and i have a problem : when in step "2g. Test TensorFlow setup to verify it works" when i run the notebook it runs till end without errors but no results , could you please help , i also open an issue in github, hope you will help thanks

  9. should I repeat all the steps everytime I want to run it ?? and it works by the way thank u

  10. training stop after some steps with error "File "C:UsersRAMAppDataLocalProgramsPythonPython37libsite-packagestensorflowpythonframeworkops.py", line 2005, in _init_

    self._traceback = tf_stack.extract_stack()"
    please Help

  11. hey can I use it with cpu ???

  12. i haven't gpu card, can I run this scripts?

  13. Hi, in the first I say thanks for your best and complete tutorials,
    I have a dataset with 652 pictures (train: 521 — test: 131) also I have 4 classes.
    In the first train, I used "faster_rcnn_inception_v2_coco" model and training reach 170000 steps and 0.04 loss. It's work perfectly and I can find all objects that I need.
    But now, I use "ssd_mobilenet_v2_coco" model and training reach 51000 steps and my loss values are larger than 1.4 and it changes very very slowly.
    additionally, pictures in dataset are 300×300 and size less than 200K.
    Is it normal that I have 1.4 loss value or I getting overfitting or underfitting ???

  14. Sorry but can anyone tell me what is this "weights" ?

  15. Quickie : Got the model to work using 2 classes , my question is , how can I pick a class and mark a co-ordinates on that class alone . Your valuable inputs are welcome . I got the co-ordinates but how can I get for only one class …

  16. I want to print the class.How can i do?

  17. By the way all, what occurred on my side was that the transormations to xml were achieved but the file was not of type xml, so you have to rename all the files with an extension '.xml' if you want the command "python xml_to_csv.py' to be successful. Otherwise you will only get the created files with headers and nothing else.

  18. Once I run "xml_to_csv.py" it creates a raccoon_labels csv with a few headers with no values in it?!?! How come? And by the way, I created a new images folder and put there 2 folders (train and test) where I split my random images and keep the xml's

  19. Can I run this on ubuntu too?

  20. what exactly going with Every Step , it seem for me i ran the code on my data-set log losses values different from time to time although all inputs still same just ran the train any one have explanation
    [lease

  21. Hi, I tried to run the training, but after around 200 steps, the loss explodes. Has anyone experienced the same problem? Thanks

  22. I can create an frozen_inference_graph with no issues but when I try to use it on my Raspberry Pi using your pet detection tutorial I get a "bus error" with my frozen inference graph. Any idea what I could be doing wrong?

Leave a comment

Your email address will not be published.


*