skyy93 / sample4geo Goto Github PK

Thanks to the author for the well-commented code. I find a mistake in code.
In the train_university.py (line94-line95)：
config.query_folder_train = './data/U1652/train/satellite' config.gallery_folder_train = './data/U1652/train/drone'
which should change to ：
config.query_folder_train = './data/U1652/train/drone' config.gallery_folder_train = './data/U1652/train/satellite'

Thanks for the great work, have you tried ConvNext-B without sharing weights?

Thanks for the great work, could you please give me the code you used to plot the inverse polar transformation in your final supplementary material?

Thanks for your response! However, the current situation is that if I train using the default parameters you suggested, the performance of the University-1652 dataset network will be poor in the Epoch=1 phase when using a 4*3090GPU.

The problem we notice is that the loss does not decrease throughout the training process.
The only difference from the default code in training is that we downloaded the ConvNeXt-T model from https://github.com/facebookresearch/ConvNeXt fine-tuned on the ImageNet-1k dataset and load it locally via
model_state_dict = torch.load('. /pretrained/university/{}.pth'.format(config.model)) model.load_state_dict(model_state_dict, strict=False)

Originally posted by @MingkunLishigure in #1 (comment)

Thank you for your excellent work, but I have some doubts. You mentioned in the ablation section of the paper that CVUSA can also use GPS sampling, but the data set does not seem to provide latitude and longitude.

Thank you very much for your paper and code, and I would like to ask you about The InfoNCE Loss
Does temperature parameter τ have an initial value? According to the thesis, it is a learnable parameter. How does it learn by itself

Hi @Skyy93 ,

I want to report your ``Random" results of CVACT_val in my paper. But I cannot find that. Can you tell me?

Best,
Guopeng.

Hi, thank you very much for your excellent work and code. Your code structure is intuitive and effective and has inspired me a lot. However, I do not have that high-performance computer in my reproduction experiments, so I tried to reduce the batch size from 128 to 16 at the expense a liitle experimental results performance. However, during the evaluation phase of the VIGOR dataset, the program often gets killed due to lack of memory, I tried to buy an extra memory stick to increase my computer's memory to 32GB, and it still gets killed. So I would appreciate if you could share the size of the memory of your experimental equipment for reference.

In addition, I noticed the following settings in the training scripts :

neighbour_select: int = 64     # max selection size from pool
neighbour_range: int = 128     # pool size for selection

Based on your description in 3.4 of the paper, while reducing the batch size to 16, should I adjust neighbour_range to 16 and neighbour_select to 8?

Hello, thank you for your great work.. We tested your method using the 'convnext_tiny.fb_in22k_ft_in1k_384' model and default parameters, training for 40 epochs. However, the model's performance only reached around 40% mean average precision on the University-1652 dataset. This suggests a deviation from the experimental results reported in your paper. Is it possible that this is due to parameter issues, such as the default setting of epoch to 1 in the training code?