[Discourse.ros.org] [Embedded] Deep Neural Network on Embedded?

[Ildoo Kim]

ildoonet September 11

I recently implement couple of algorithms for Deep Learning, which provides fast inference on an embedded devices.

One of them is called ‘LCNN’, you can find the original paper here : https://arxiv.org/abs/1611.06473

And I implemented it with tensorflow,

github.com

ildoonet/tf-lcnn

tf-lcnn - Tensorflow implementation for 'LCNN: Lookup-based Convolutional Neural Network'. Predict Faster using Models Trained Fast with Multi-GPUs

This codes compress alexnet which takes roughly 150ms or more on a single core cpu,

to a sparse convolutional layered network which takes 10~50ms on the same environment.

https://github.com/ildoonet/tf-lcnn/raw/master/images/timeline_alexnet.png

So… based on these new technologies, i am looking for an idea to try.

Like openpose on a robot : Human Pose Estimation Deep Learning Model (OpenPose) ROS Package

Any Thoughts?

---

Visit Topic or reply to this email to respond.

To unsubscribe from these emails, click here.

[Víctor Mayoral Vilches]

vmayoral September 11

@ildoonet, that’s pretty cool. What embedded platform are you working on? The results are pretty interesting. Well done.

[Ildoo Kim]

ildoonet September 14

It’s for snapdragon cpus like smartphones.

Also I’m working on arm cpus.

Also, this codes is not optimized for specific cpu architecture, It can be improved further.

---

Visit Topic or reply to this email to respond.

---

In Reply To

vmayoral September 11

[David Crawley]

davecrawley September 19

That’s really cool!

Do you have trained neural nets to go along with it?

[Ildoo Kim]

ildoonet September 19

I have alexnet which 20 times faster than the original one. I will add more models. Do you have any particular model you want?

---

Visit Topic or reply to this email to respond.

---

In Reply To