Joshua Z. Zhang

1. I haven't got time to setup the CI/CD so it's not forced to check the unittests 2. The GetFrameCount now relies on frame_ts_ strictly after the initial check from...

Can you guys post your cuda version/ decive type? I have tried on my local machine with 1070ti and cuda 10.1.243, didn't notice any mem leak. ```python from decord import...

Can you guys post the kaggle gpu and cuda version?

What's your cmake version? @YuqiHUO

See #17

I don't think I understand your question. If you have images why not just feed in the image?

You can loop through the `VideoReader` and pass it to display with the numpy array.

@yztongzhan I will take a look first

@yztongzhan it appears that your gcc version is `5.4.0`, I am not sure if it's relevant but please pull the latest master with: e676a688e6c7fd49797866102d3388fe72b5d4fd

@douglas125 Have you tried setting the output format to be tensorflow? For example ```python import decord decord.bridge.set_bridge('tensorflow') ```