rvandeghen

@datumbox First, I rather use what I call an InfiniteSampler, defined as ``` class InfiniteSampler(torch.utils.data.Sampler): def __init__(self, sampler): self.sampler = sampler def __len__(self): return len(self.sampler) def __iter__(self,): while True: yield...

If you don't use `retinanet = torch.nn.DataParallel(retinanet).cuda()` you should write `retinanet.freeze_bn()` instead of `retinanet.module.freeze_bn()`. Thus to solve all the issues, make sure that all your call with retinanet don't use...

@JanuszL thanks for the reply. Do you know how to return both list of frames and list of OF ? I have an error which I guess comes from the...

Hi @JanuszL, Indeed the optical flow gives good results at barely no extra cost. However, I found in the [blogpost](https://developer.nvidia.com/blog/an-introduction-to-the-nvidia-optical-flow-sdk/) the following information and I would like to know if...

Hi @JanuszL, Do you know if I should expect huge/small changes in the output depending on the value I set to `hint_grid` ? I did some comparisons between the NVIDIA...

Hi @stiepan, I'm using pytorch, and I would like to have data coming from videos and from images. Currently, I was using them separately, so I have 2 independent pipes...

@stiepan Here are some code snippets that we currently use. For the images, we have a pipeline something similar to ``` @pipeline_def def create_image_reader_pipeline(files, shard_id=0, num_shards=1): files, _ = fn.readers.file(files=files,...

@szalpal Thanks for your fast response. Yes you are right, and since I'm new to DALI, I try to correctly understand the API. Because I have different length duration, I...

@szalpal, Just to let you know, I'm already able to use DALI in the case of training. However, I was also concerned to use DALI to process my video files...

@JanuszL, Thanks for the feedback ! I had a chat with someone who created the videos and he is aware that for some of them, he also experienced a problem...