Generating Interpretable Poverty Maps using Object Detection in Satellite Images
Authors: Kumar Ayush, Burak Uzkent, Marshall Burke, David Lobell, Stefano Ermon
IJCAI 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | 6 Experiments and Table 2: LSMS poverty score prediction results in Pearson s r2 using parent level features (YOLOv3 trained on 10 classes) and child level features (YOLOv3 trained on 60 classes). |
| Researcher Affiliation | Academia | Kumar Ayush1 , Burak Uzkent1 , Marshall Burke2 , David Lobell2 and Stefano Ermon1 1Department of Computer Science, Stanford University 2Department of Earth System Science, Stanford University {kayush, buzkent}@cs.stanford.edu, mburke@stanford.edu, dlobell@stanford.edu, ermon@cs.stanford.edu |
| Pseudocode | No | The paper describes its methods verbally in the text but does not include any explicitly labeled pseudocode or algorithm blocks. |
| Open Source Code | No | The paper does not contain an explicit statement that the authors' implementation code for the described methodology is open-source, nor does it provide a link to a code repository. |
| Open Datasets | Yes | We use x View [Lam et al., 2018], as our source dataset. It is one of the largest and most diverse publicly available overhead imagery datasets for object detection. The dataset comes from field Living Standards Measurement Study (LSMS) survey conducted in Uganda by the Uganda Bureau of Statistics between 2011 and 2012 [UBOS, 2012]. |
| Dataset Splits | Yes | Due to small size of the dataset, we use a Leave-one-out cross validation (LOOCV) strategy. Since nearby clusters could have some geographic overlap, we remove clusters which are overlapping with the test cluster from the train split to avoid leaking information to the test point. |
| Hardware Specification | Yes | We perform object detection on 320 1156 9 chips (more than 3 million images), which takes about a day and a half using 4 NVIDIA 1080Ti GPUs. |
| Software Dependencies | No | The paper mentions using YOLOv3 and DarkNet53 but does not provide specific version numbers for these or any other software dependencies. |
| Experiment Setup | Yes | YOLOv3 is usually used with an input image size of 416 416 px. We fine-tune the weights of the YOLOv3 model, pre-trained on the Image Net, using the training split of the x View dataset. The detections are obtained using a 0.6 confidence threshold. |