Towards Domain Adaptive Vehicle Detection in Satellite Image by Supervised Super-Resolution Transfer
Authors: Liujuan Cao, Rongrong Ji, Cheng Wang, Jonathan Li
AAAI 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | With comprehensive experiments on large-scale satellite image collections, we demonstrate that the proposed framework can significantly boost the detection accuracy over several state-of-the-arts. |
| Researcher Affiliation | Academia | Liujuan Cao , , Rongrong Ji , , Cheng Wang , , Jonathan Li , Fujian Key Laboratory of Sensing and Computing for Smart City, Xiamen University, 361005, China School of Information Science and Engineering, Xiamen University, 361005, China {caoliujuan,rrji,cwang,junli}@xmu.edu.cn |
| Pseudocode | Yes | Algorithm 1: Domain-Adaptive Vehicle Detection in Satellite Images by Super Resolution Transfer |
| Open Source Code | No | The paper does not provide any statement or link indicating that the source code for the described methodology is publicly available. |
| Open Datasets | No | To build the satellite image dataset, we collect 80 satellite images from Google Earth, in which each image is with 979 1348 resolution, covering the road maps in New York City. Correspondingly, we further collect 80 corresponding aerial images covering the same road map of New York City by zooming in the Google Earth into the finest resolution. We ask a group of volunteers to manually labeled vehicle regions with both lowand high-resolution images collected above, which produces 1,482 vehicle annotations in total. |
| Dataset Splits | Yes | From the aerial images, we adopt a 1:5 leave-oneout (training vs. validation) setting for parameter tuning. Note that labels from the low-resolution satellite images are used for validation purpose only. |
| Hardware Specification | No | The paper does not provide specific details about the hardware (e.g., GPU/CPU models, memory) used for running the experiments. |
| Software Dependencies | No | The paper mentions software like 'Map Info/SHP format 2D-vector map and Arc GIS Engine', and specific algorithms, but does not provide version numbers for any key software components or libraries. |
| Experiment Setup | No | The paper discusses parameter tuning for dictionary size, percentage of supervised labels, and penalty C in E-SVMs, but does not explicitly provide the final specific hyperparameter values or detailed system-level training settings used for the experiments. |