graviti 0.2.0

Graviti TensorBay python SDK

Graviti python SDK

Installation

pip3 install graviti

Usage

Get accessKey

AccessKey is required when upload data.

Use your username and password to login to Graviti website, and get accessKey on profile page.

Upload data

This sample is for uploading dataset which only contains data collected from a single sensor.

#!/usr/bin/env python3

from graviti import GAS

ACCESS_KEY = "Accesskey-****"
DATASET_NAME = "TestDataset"
SEGMENT_NAME = "segment1"

gas = GAS(ACCESS_KEY)

dataset = gas.get_or_create_dataset(DATASET_NAME)  # create dataset
segment = dataset.get_or_create_segment(SEGMENT_NAME)  # create segment

# single-thread upload
for filename in FILE_LIST:
    segment.upload_data(filename, "remote_path")
# 'remote_path' should follow linux style. If it has wrong format, raise GASPathError.

# # multi-thread uploading
# # when use multi-thread uploading
# # uncomment the following code and comment single-thread uploading code

# from concurrent.futures import ThreadPoolExecutor

# THREAD_NUM = 8 # number of threads

# with ThreadPoolExecutor(THREAD_NUM) as executor:
#     for filename in FILE_LIST:
#         executor.submit(segment.upload_data, filename, "remote_path")

Upload fusion data

This sample is for uploading dataset which contains data collected from fusion.

#!/usr/bin/env python3

from graviti import GAS
from graviti.dataset import Frame, Data
from graviti.sensor import Camera

ACCESS_KEY = "Accesskey-****"
DATASET_NAME = "TestFusionDataset"
SEGMENT_NAME = "segment1"

gas = GAS(ACCESS_KEY)

dataset = gas.get_or_create_dataset(DATASET_NAME, is_fusion=True)  # create fusion dataset
segment = dataset.get_or_create_segment(SEGMENT_NAME)  # create segment

for sensor_name in SENSOR_LIST:
    camera = Camera(sensor_name)
    camera.description = "This is a camera"
    camera.set_translation(x=1.1, y=2.2, z=3.3)
    camera.set_rotation(w=1.1, x=2.2, y=3.3, z=4.4)
    camera.set_camera_matrix(fx=1.1, fy=2.2, cx=3.3, cy=4.4)
    camera.set_distortion_coefficient(p1=1.1, p2=2.2, k1=3.3, k2=4.4, k3=5.5)

    segment.upload_sensor_object(camera)

# single-thread uploading
for index, frame_info in enumerate(FRAME_LIST):
    frame = Frame()
    for sensor_name in SENSOR_LIST:
        frame[sensor_name] = Data(
            frame_info.filename, remote_path="remote_path", timestamp=frame_info.timestamp,
        )
    # 'remote_path' should follow linux style. If it has wrong format, raise GASPathError.
    segment.upload_frame_object(frame, index)

# # multi-thread uploading
# # when use multi-thread uploading
# # uncomment the following code and comment single-thread uploading code

# from concurrent.futures import ThreadPoolExecutor

# THREAD_NUM = 8  # number of threads

# with ThreadPoolExecutor(THREAD_NUM) as executor:
# for index, frame_info in enumerate(FRAME_LIST):
#     frame = Frame()
#     for sensor_name in SENSOR_LIST:
#         frame[sensor_name] = Data(
#             frame_info.filename, remote_path="remote_path", timestamp=frame_info.timestamp,
#         )
#         executor.submit(segment.upload_frame_object, frame, index)

Command line

We also provide gas command to call SDK APIs.

Use gas in terminal to see the available commands as follows.

gas config
gas create
gas delete
gas publish
gas ls
gas cp
gas rm

config environment

gas config [accessKey]                   # config accesskey to default environment
gas -c [config_name] config [accessKey]  # create an environment named [config_name]
                                         # and config accesskey to it

show config

gas config         # show config information of all environments

choose environment

gas [command] [args]                           # choose default environment
gas -c [config_name] [command] [args]          # choose the environment named [config_name]
gas -k [accessKey] [command] [args]            # appoint accessKey in current command line

# '-k' has higher priority than '-c'

create dataset

gas create tb:[dataset_name]

delete dataset

gas delete tb:[dataset_name]

publish dataset

gas publish tb:[dataset_name]

list data

gas ls [Options] [tbrn]

Options:
  -a, --all      List all files under all segments. Only works when [tbrn] is tb:[dataset_name].

tbrn:
  None                                              # list all dataset names
  tb:[dataset_name]                                 # list all segment names under the dataset
  tb:[dataset_name]:[segment_name]                  # list all files under the segment
  tb:[dataset_name]:[segment_name]://[remote_path]  # list files under the remote path

upload data

gas cp [Options] [local_path1] [local_path2]... [tbrn]

Options:
  -r, --recursive     Copy directories recursively.
  -j, --jobs INTEGER  The number of threads.

tbrn:
  tb:[dataset_name]:[segment_name]                  # copy files to the segment
  tb:[dataset_name]:[segment_name]://[remote_path]  # copy files to the remote path

# [segment_name] is must required.
# If only upload one file and [remote_path] doesn't end with '/',
# then the file will be uploaded and renamed as [remote_path]

delete data

gas rm [Options] [tbrn]

Options:
  -r, --recursive  Remove directories recursively.
  -f, --force      Delete segments forcibly regardless of the nature of the dataset.
                   By default, only segments with no sensor can be deleted.
                   Once '-f' is used, sensors along with their objects will also be deleted.

tbrn:
  tb:[dataset_name]                                 # remove all segments under the dataset
  tb:[dataset_name]:[segment_name]                  # remove a segment
  tb:[dataset_name]:[segment_name]://[remote_path]  # remove files under the remote path