library(tfdatasets)
# Creates a dataset that reads all of the examples from two files, and extracts
# the image and label features.
filenames <- c("/var/data/file1.tfrecord", "/var/data/file2.tfrecord")
dataset <- tfrecord_dataset(filenames) %>%
dataset_map(function(example_proto) {
features <- list(
image = tf$FixedLenFeature(shape(), tf$string, default_value = ""),
label = tf$FixedLenFeature(shape(), tf$int32, default_value = 0L)
)
tf$parse_single_example(example_proto, features)
}) tfrecord_dataset
A dataset comprising records from one or more TFRecord files.
Description
A dataset comprising records from one or more TFRecord files.
Usage
tfrecord_dataset(
filenames,
compression_type = NULL,
buffer_size = NULL,
num_parallel_reads = NULL
) Arguments
| Arguments | Description |
|---|---|
| filenames | String(s) specifying one or more filenames |
| compression_type | A string, one of: NULL (no compression), "ZLIB", or "GZIP". |
| buffer_size | An integer representing the number of bytes in the read buffer. (0 means no buffering). |
| num_parallel_reads | An integer representing the number of files to read in parallel. Defaults to reading files sequentially. |
Details
If the dataset encodes a set of TFExample instances, then they can be decoded into named records using the dataset_map() function (see example below).