search

vast-community-hub / tensorflow-vast

TensorFlow binding library for VA Smalltalk
MIT License
13 stars 0 forks source link

Add a simple object detection example #5

Open marianopeck opened 5 years ago

marianopeck commented 5 years ago

Now that we got image recognition to work, we shoud try "Object Detection". Something like this or this or this.

I wonder if we can download pre-trained models with that and then display the images from Smalltalk. Did you ever try? @SergeStinckwich ?

marianopeck commented 5 years ago

There are some opinions already about this:

https://github.com/vasmalltalk/tensorflow-vast/issues/4#issuecomment-518448398

https://github.com/vasmalltalk/tensorflow-vast/issues/4#issuecomment-518484149

https://github.com/vasmalltalk/tensorflow-vast/issues/4#issuecomment-518485168

gerasdf commented 5 years ago

I've been looking at the "TensorFlow Object Detection Zoo" (https://github.com/tensorflow/models/blob/master/research/object_detection/g3doc/detection_model_zoo.md) and it's absolutely possible (and somehow easy) to implement all of them...

I just need to have some time. The basic structure is the same as for LabelImage examples, so we may want to add a common superclass that loads the graph and handles a number of image files. Then what's different is the input node (we can search it by name, it's easy) and the output tensors and labels file format... We'll see... I'm swamped with work, so I can't really sit down and do it, but maybe... just maybe...

SergeStinckwich commented 5 years ago

If we split the effort we can do most of the example in the webpage I guess.