search

bourdakos1 / Object-Detection

Leverage TensorFlow's object detection with Watson Visual Recognition
14 stars 17 forks source link

invalid-api-key #3

Open GianlucaCerilli opened 5 years ago

GianlucaCerilli commented 5 years ago

Hi Nick,

very nice tutorial here https://medium.com/ibm-watson/dont-miss-your-target-object-detection-with-tensorflow-and-watson-488e24226ef3 ! But when I try to run the ¨python3 object_detection.py¨ I get the following error:

Downloading model... (This may take over 5 minutes) Model already downloaded Loading model... detecting... Traceback (most recent call last): File "object_detection.py", line 108, in results = visual_recognition.classify(images_file=images_file, parameters=parameters) File "/home/gianluca/venv/lib/python3.5/site-packages/watson_developer_cloud/visual_recognition_v3.py", line 119, in classify accept_json=True) File "/home/gianluca/venv/lib/python3.5/site-packages/watson_developer_cloud/watson_service.py", line 409, in request error_info) watson_developer_cloud.watson_service.WatsonApiException: Error: invalid-api-key, Code: 403

Do you have any idea about it? Thank you very much!

bourdakos1 commented 5 years ago

ah sorry, this repo is a bit dated. They changed the name of the api variable name to something like iam_apikey. I can look at the docs and update the repo

GianlucaCerilli commented 5 years ago

Thank you very much for the immediate reply and for the availability. I am following your really nice article "Custom Object Detection with Tensorflow". Is there something dated too in that article or is it possible to follow it? Do you also know if it's possible to use the final script (object tracking) in ROS kinetic?

Thank you very much again!

GianlucaCerilli commented 5 years ago

Hi @bourdakos1 , can you tell me something about my last message?

Thank you very much!

bourdakos1 commented 5 years ago

Sorry I didn’t see this until now. I don’t know much about ROS kinetic so I’m not sure if this is compatible or not. Also the article should be fairly up to date. The only part that is out of date is the Watson api. The Watson api has been changing a lot recently and doesn’t play a large roll in the results, so you could probably remove it from the project.

GianlucaCerilli commented 5 years ago

HI @bourdakos1,

I am running now the training (800 images with a NVIDIA Quadrop2000).

After having changed the xmls , trainval and labelmap files, the only other things to change are the values inside faster_rcnn_resnet101.config ? I am using this config as you, but with only one custom object to detect. So, I have only changed num_classes from 2 to 1. Are there other changes to do?

Thank you very much again