I’m using a custom data layer provided in a fork to read video frames. This implementation reads images from a directory using OpenCV. Now, I’m modifying this layer to read from HDF5 files instead (because of compression and to avoid exceeding #inodes). However, I’m struggling because the original layer uses Datum
, while the Caffe HDF5 loaders output Blob
.
I have already done most of it using Blob
, but I still have to implement a few functionalities. The original layer provides an option to resize the images using cv::resize
. Also, it reads a batch of images and concatenates them along the channel axis, generating a single blob of images.
I’m implementing a custom data layer to read batches of frames from HDF5 files for video processing. I was wondering if there are any functions or examples available that can help me with any of the following:
concat_layer.cpp
does it)cv::resize
?Any tips are welcome! =)