Subclass of threedb.result_logging.base_logger.BaseLogger.

class threedb.result_logging.tb_logger.TbLogger(tb_dir: str, result_buffer: threedb.utils.CyclicBuffer, _)

Bases: threedb.result_logging.base_logger.BaseLogger

TensorBoard Logger.

Logs accuracy (as a scalar) and all generated images to TensorBoard.

__init__(tb_dir: str, result_buffer: threedb.utils.CyclicBuffer, _)None

Creates a logger instance

  • root_dir (str) – The directory in which to write the logging results (should be the same for all loggers, with each logger making a subfolder to log in)

  • result_buffer (CyclicBuffer) – The buffer where the main thread is writing the results

  • config (Optional[Dict[str, Dict[str, Any]]]) – The config file (parsed from YAML) of the 3DB experiment being run


Abstract method for logging an item from the buffer


item (Dict[str, Any]) – A dictionary containing the results of a single rendering (as returned by threedb.client).


Performs cleanup operations for the logger. No-op by default, should be overriden with code for closing any open file handles, ports, etc.


alias of threedb.result_logging.tb_logger.TbLogger