This instruction contains details about how to convert the VisualGenome dataset into a format that can be read by the framework. Alternatively,you may follow the instruction to download a pre-processed dataset.
Description PolyUmultispectral palmprint Database: Multispectral palmprint. Multispectral Palmprint Database: MSpalmprint Database.zip ROI. RoI distribution: bboxdistribution.npy mini-vg: Alternatively, you may download a subset of the dataset (mini-vg). The dataset contains 1000 images (973 have scene graph annotations) from the test set of the full VG dataset and no training data. Mini-vg takes around 4GB of space uncompressed.
Overview
A dataset for the framework consists of five files:
- An image database in hdf5 format.
- An scene graph database in hdf5 format.
- An scene graph database metadata file in json format.
- An RoI proposal database in hdf5 format.
- An RoI distribution for normalizing the bounding boxes.
Important: Note that (1) takes ~320GB of space. Hence we recommend creating (1) by yourself and download (2-5). If you just want to test/visualize predictions,you may download a subset of the processed dataset following the instructions in the next section.Also note that the framework does not include a regional proposal network implementation. Hence (4) is needed to run the framework.
Download pre-processed dataset.
You may download the pre-processd full VG dataset using the following links
- Image database (currently unavailable, please refer to the next section on how to create an IMDB by yourself)
- Scene graph database: VG-SGG.h5
- Scene graph database metadata: VG-SGG-dicts.json
- RoI proposals: proposals.h5
- RoI distribution: bbox_distribution.npy
mini-vg: Alternatively, you may download a subset of the dataset (mini-vg). The dataset contains 1000 images (973 have scene graph annotations) from the test setof the full VG dataset and no training data. mini-vg takes around 4GB of space uncompressed. Note that you don't have to download it if you have ran the
download.sh
script.After downloading the dataset, place all files under
data/vg/
. If you use the mini-vg dataset, the directory should look like this:Convert VisualGenome to desired format
(i). To start with, download VisualGenome dataset using the following links:
- Images part1part2
(ii). Place all the json files under
data_tools/VG/
. Place the images under data_tools/VG/images
(iii). Create image database by executing
./create_imdb.sh
in this directory. This script creates a hdf5 databse of images imdb_1024.h5
.The longer dimension of an image is resized to 1024 pixels and the shorter side is scaled accordingly. You may also create a image database of smaller dimension byediting the size argument of the script. You may skip to (vii) if you chose to downloaded (2-4).(iv). Create an ROI database and its metadata by executing
./create_roidb.sh
in this directory. The scripts creates a scene graph database file VG-SGG.h5
and its metadata VG-SGG-dicts.json
.By default, the script reads the dimensions of the images from the imdb file created in (iii). If your imdb file is of different size than 512 and 1024, you must add the size tothe img_long_sizes
list variable in the vg_to_roidb.py
script.(v). Use the script provided by py-faster-rcnn to generate (4).
(vi). Change line 93 of
tools/train_net.py
to True
to generate (5).(vii). Finally, place (1-5) in
data/vg
.A postal code (known in various countries as a post code , postcode , or ZIP code ) is a series of and/or digits appended to a postal address for the purpose of sorting mail.Postal codes are codes designating the areas that each post office serves. So it depends on where in Lebanon your letter is addressed. ( code postal liban )
Address format for Lebnon
Mr. Hassan
526 bina Street. 2nd floor
RAOUCHEH BEIRUT 2038 3054
LEBANON
526 bina Street. 2nd floor
RAOUCHEH BEIRUT 2038 3054
LEBANON
Postal Codes in Lebanon consist of 4 numbers,also consist of 8 numbers
Example : 2 0 3 8 3 0 5 4
The first four digits represent the region or postal zone,the last four digits represent the building.
Follows this links of all post offices in Lebanon, their address and their postal code.
Beirut District
Mount Lebanon 1
Mount Lebanon 2
South Lebanon
North Lebanon
Bekaa
Mount Lebanon 1
Mount Lebanon 2
South Lebanon
North Lebanon
Bekaa