Hi @ccmdi,
I wonder in all of your benchmarks shown at https://geobench.org/ (ACW, AVW, Rural, Urban, Photos), did you always give the same set of examples for all models, i.e. same image as input and same GT latitude/longitude/country for evaluation? How many examples in total? 100?
If I want to generate an evaluation set myself, should I just do python dataset.py --num 100 --output ACW --map 62a44b22040f04bd36e8a914 for ACW? How about AVW, Rural, Urban, and Photos? Could you please send the commands so that I can also create the data for those?
Thanks so much again in advance!!
Hi @ccmdi,
I wonder in all of your benchmarks shown at https://geobench.org/ (ACW, AVW, Rural, Urban, Photos), did you always give the same set of examples for all models, i.e. same image as input and same GT latitude/longitude/country for evaluation? How many examples in total? 100?
If I want to generate an evaluation set myself, should I just do
python dataset.py --num 100 --output ACW --map 62a44b22040f04bd36e8a914for ACW? How about AVW, Rural, Urban, and Photos? Could you please send the commands so that I can also create the data for those?Thanks so much again in advance!!