AI for City Districts: Measuring City Districts’ Sustainability using Satellite Data

This repository contains codes and data for performing city index prediction (regression) task.

Original dataset contains raw 1302 images of 4800 x 4800 resolution (14 GB).

• Raw data and processed dataset can be downloaded at our Hugging Face dataset link.
• Models (both pre-trained, and fine-tuned) can be accessed at Hugging Face models link.

Dataset consists of 45 cities from various locations, and mostly chosen from Arcadis Index 2022. Cities and their corrresponding index values from 9 different sustainability ranking systems which were used for model training:

City IATA Code	Overall Arcadis SCI	Planet Arcadis SCI	People Arcadis SCI	Profit Arcadis SCI	Sustainable cities by Corporate Knights	Resilient Cities by Grosvenor	Global Cities by AT Kearney	European Green City Index	US and Canada Green City Index
ALA	-	-	-	-	-	-	118	-	-
ESB	-	-	-	-	-	-	86	-	-
NQZ	-	-	-	-	-	-	128	-	-
GYD	-	-	-	-	-	-	-	-	-
BKK	72	92	58	73	-	-	35	-	-
PEK	73	91	71	53	30	39	6	-	-
FRU	-	-	-	-	-	-	-	-	-
BOG	78	20	82	82	36	-	63	-	-
BOS	22	54	54	3	-	7	21	-	6
BNE	64	60	57	50	-	27	-	-	-
AEP	82	62	84	84	35	36	32	-	-
CAI	86	89	79	91	-	48	59	-	-
CHI	52	67	70	8	-		8	-	11
DUB	37	28	19	65	-	29	45	21	-
HAN	85	93	80	85	-	-	-	-	-
HKG	63	56	65	45	-	30	7	-	-
IST	74	55	74	79	46	-	27	25	-
CGK	83	68	81	86	-	49	67	-	-
FIH	100	99	95	100	-	-	136	-	-
KUL	71	73	62	69	-	-	-	-	-
LOS	99	88	100	99	40	-	113	-	-
LHE	94	95	85	97	-	-	127	-	-
LIS	57	24	56	66	-	-	46	18	-
MNL	93	83	97	89	-	47	69	-	-
MEL	60	50	61	43	-	-	12	-	-
MEX	79	53	83	77	-	44	31	-	-
MIL	51	21	39	71	-	33	44	-	-
BOM	91	81	89	96	44	46	62	-	-
MUC	19	12	25	27	-	24	26	-	-
NBO	96	82	98	95	-	-	89	-	-
OSL	1	1	17	39	2	-	54	3	-
PAR	8	2	43	31	17	23	3	10	-
RIX	44	18	30	62	-	-	-	15	-
SFO	9	35	38	4	16	16	11	-	1
GRU	84	44	94	78	42	41	40	-	-
ICN	26	43	4	44	25	35	17	-	-
CIT	-	-	-	-	-	-	-	-	-
SIN	35	69	5	28	45	32	9	-	-
SYD	33	42	15	46	26	19	15	-	-
TPE	46	71	20	29	-	34	49	-	-
TAS	-	-	-	-	-	-	-	-	-
TKY	2	7	7	20	-	26	4	-	-
YVR	17	26	13	30	8	2	39	-	2
IAD	20	45	37	15	24	9	14	-	8

📥 Downloading the repository

git clone https://github.com/IS2AI/city-classification-and-index-prediction

⚙️ Pre-processing

Prior to training it is necessary to perform pre-processing on raw images. To generate patches out of raw images needed for training, and to perform train-val-test split launch the following script:

python3 preprocessing.py

⚡ Training

To launch training for city index predition use:

python3 train_regression.py

🎯 Inference

To test city index prediciton on unseen patches run:

python3 test_regression.py

🌱🗺️ Sustainability map creation

To create sustainability color map, there is available another script:

python3 make_sustainability_map.py

🔍🤖 Explainable AI

To run Relevance-CAM and receive Vizual Explanations of decision making process of models, run:

python3 rel_cam.py

Full list of Vizual Explanations produced using Relevance-CAM can be found here