Object Detection with Open Images using Tensorflow

One thing that really puts me off Tensorflow is the large amount of work you have to do just to get your data in.<p>I recently trained a detector using Nvidia&#x27;s detectnet and the DIGITS environment really makes it easy from dataset&#x2F;database creation to model generation and training. The web interface handles all the padding and resizing for you when you import your images, and detectnet does online augmentation during training. It takes data in KITTI format which is just image + textfile with labels. No wrangling in Python needed.<p>The documentation is pretty terrible though, you have to pore over github issues to fix and modify things, but it does work.

Author here, if you have any questions&#x2F;comments I&#x27;ll check back in periodically and answer them here!

I mean, this is neat but as someone who actually tried to build a computer vision product, can I just say Open Images data aren&#x27;t quite enough? Also, computer vision isn&#x27;t quite at &quot;human level&quot; yet. For your own project, building a model that has 90% accuracy on the test set is awesome but for an actual product to be released into the wild, it could have serious problems (not to mention adversarial examples).

A minor nit but the function to dedup the image ids and the corresponding comments seem to be off from a data structure&#x2F;algorithmic POV.<p>&quot;Looking at our deduplication function, it’s functional and performant, but not very descriptive. Essentially it checks a running set called seen, which is checked for originality as the deduplication script progresses. As python Dicts are essentially a hash map, the in check compares element hashes instead of each individual component of the dictionary object. This massively speeding up the deduplication process.&quot;<p>Uhm..I don&#x27;t think so.<p>It is just extra work to check if the objects are already in the set vs just stuffing them all in and let the set handle uniqueness.<p>Am I missing something?<p><a href="https:&#x2F;&#x2F;gist.github.com&#x2F;pstoll&#x2F;ae73582763540051d321a4eb15304226" rel="nofollow">https:&#x2F;&#x2F;gist.github.com&#x2F;pstoll&#x2F;ae73582763540051d321a4eb15304...</a><p>Again, a minor thing. But seeing something like that makes my &#x27;what else do I need to review&#x27; detectors go up.

live demo: <a href="https:&#x2F;&#x2F;algorithmia.com&#x2F;algorithms&#x2F;zeryx&#x2F;openimagesDemo" rel="nofollow">https:&#x2F;&#x2F;algorithmia.com&#x2F;algorithms&#x2F;zeryx&#x2F;openimagesDemo</a>