I'm in a group project in school and we are currently using the tensorflow object detection API. The object detection works great but we are very interested in how the developers of this API have tested it. Is there anyone who has contributed to the project or knows where I can find more information about testing?
Yes, the API is well tested.
You can find the tests of each module in a python file at the same level and with the same name + the suffix "_test"
As an example the module:
https://github.com/tensorflow/models/blob/master/research/object_detection/model_lib.py
Is tested in:
https://github.com/tensorflow/models/blob/master/research/object_detection/model_lib_test.py
Related
No example of this api is given in the TensorFlow documentation. And also I didn't find any code snippet related to this api elsewhere. So I use ChatGPT to generate some code snippet, but they didn't work wellðŸ˜.
Has anyone ever called this api successfully, if so, could you please provide some code snippets for my reference? And the version of tensorflow is 2.0+.
I am struggling to find the API for a particular class in tensorflow.
For example, I am interested in finding the specs for this class:
tensorflow.python.framework.ops.EagerTensor
Sorry if this sounds as a stupid question.
Tensorflow is not that well documented yet. The docs are generated from the comments (the project is hosted at GitHub) and in your case, there are none.
For insights like that, you will have to read the source code.
tf.nn.seq2seq.embedding_attention_seq2seq
I was searching the above function in tensorflow api. But I counln't find that by the exact name. But similar function was there in the class tf.contrib.legacy_seq2seq.embedding_attention_seq2seq .
Why they moved few from nn class to contrib class ?
Google just released TensorFlow 1.0 last week, where the API has been updated, so things have been moved around. Because of the various contributions from the TF community, the contrib directory is reserved for this code, which has not yet been finalized for acceptance into core TF.
Check out the README on contrib:
The contrib directory contains project directories, each of which has
designated owners. It is meant to contain features and contributions
that eventually should get merged into core TensorFlow, but whose
interfaces may still change, or which require some testing to see
whether they can find broader acceptance. We are trying to keep
duplication within contrib to a minimum, so you may be asked to
refactor code in contrib to use some feature inside core or in another
project in contrib rather than reimplementing the feature.
I want to add an API like git_merge_octopus_bases, but how perform the local test cases? Could please show me some documents for this kind topics.
The libgit2 project relies on the Clar test framework.
You can get a first grasp of Clar features and usage in the Readme file.
The following documents should help one understand the basics of contributing to libgit2:
Contributing guidelines
LibGit2 conventions
Since I haven't got any response on the Unity3d or Evernote forums, I'll try it here.
The last year I have worked a lot with Unity3D, mostly because the good integration with the Vuforia Augmented Reality library and the fact that publishing for multiple platforms is a piece of cake.
Now I want to show notes in an AR setting and am looking at the Evernote API for this. I couldn't find anything about using this with Unity, I can see why this is not the most common combination.
My question is: do you think I can access the Evernote API through Unity? If so, how should I do this? Or is it for this purpose perhaps wiser to make (parts of) the application with Eclipse/xCode?
Hope to hear from you!
Link to Evernote API: http://dev.evernote.com/doc/
The Evernote API has a C# SDK which you should be able to call through Unity. In terms of how to do it, you will probably need to download the SDK and follow the instructions yourself. Their github seems like a good starting point.
One thing to note is that Unity's .Net library for mobile clients are quite limited and with webplayer you will need to deal with sandbox security issues. But start with the standalone build first and see how you go