Request: DOI
i need the dataset to re-train and test new models can you please provide the dataset
also please provide the architecture of the model.
Hello, I went looking for the original dataset of this model, but couldn't find it. If I remember correctly it was a small dataset compiled from still frames of online drone videos after flooding. The architecture is VGG16.
can you please tell where i can find the images i want to train the model with some changes.
I don't remember exactly. It was part of a hack-a-thon project originally so parts of it haven't been well preserved. I believe most of the still frame samples were taken from a few youtube videos though (see something like this https://youtu.be/dB05wYRamQA). From there the original dataset of ~50 hand-selected images was augmented using SDXL to ~1000 and other drone footage of cities and countrysides were used as negatives.
if you got the project files can you share it with me.