Alexander Swerdlow

Were you able to fully compile the model for the edge TPU? I was only able to get 39 supported ops + 8 that will run on the CPU for...

@hhk7734 Yep. I’ll double check tomorrow but I’m almost positive. The only significant change I made was leakyRelu -> Relu. I was able to get it running on the edge...

Can't share the tflite but happy to share my config. It's basically the stock config just without leakyRelu, 416x416, the right filter value, etc. [yolo-tiny-v4-obj.txt](https://github.com/hunglc007/tensorflow-yolov4-tflite/files/5139041/yolo-tiny-v4-obj.txt) Edit if you're curious, here's...

I have the exact same problem with the same number/type of unsupported ops. Also on tf 2.3.0 (gpu).

Update: Tensorflow 2.1.0 seems to work for me. Didn't have to change anything about the EdgeTPU compiler, just the TF version while running keras_to_tflite_quant.py

We have this issue. Closing shuffleboard after every robot code restart is possible but very frustrating. After a couple minutes, shuffleboard takes 100% of cpu usage to the point where...

@edgarriba I hate to say this but I don't think I'll have time to dig into this more in the very near future. From what it sounds like, my implementation...

Upon digging a little deeper, I think I figured out the issue which is pretty simple albeit unintuitive for an end-user. Keypoints can be transformed to be moved out of...

I don't think it should be very difficult but I'm not very familiar with Kornia (this is my first time using it actually). I added a [basic implementation](https://github.com/kornia/kornia/pull/2689) but I'm...

Also experiencing this issue using Stage 2 training w/PyTorch Lightning. Missing a large number of keys when converting back from the deepspeed checkpoint though I can see them when using...