Thank for your significant work! And could you provide the pre-trained model on sysu?

Hello code owner, I would like to ask you whether this code is the code of the original paper or the code that reproduces the original paper? Are the training results much different from the original paper? How many mAPs are there?

我的是40左右

When I act python train.py --dataset sysu

==========
Args:Namespace(arch='resnet50', batch_size=4, dataset='sysu', delta=0.2, gpu='1', img_h=384, img_w=192, log_path='log/', lr=0.1, margin=0.3, method='agw', mode='all', model_path='save_model/', num_pos=4, optim='sgd', resume='', save_epoch=20, seed=0, test_batch=64, test_only=False, trial=1, vis_log_path='log/vis_log/', workers=4)

==> Loading data..
Traceback (most recent call last):
File "train.py", line 133, in
gall_img, gall_label, gall_cam = process_gallery_sysu(data_path, mode=args.mode, trial=1)
File "/home/MMN/data_manager.py", line 58, in process_gallery_sysu
files_rgb.append(random.choice(new_files))
File "/home/anaconda3/envs/lib/python3.7/random.py", line 261, in choice
raise IndexError('Cannot choose from an empty sequence') from None
IndexError: Cannot choose from an empty sequence

Please ask where it might go wrong？

ValueError: Required crop size (384, 192) is larger then input image size (404, 164) 我在运行python train.py --dataset sysu 直接将384, 192改成384, 144可以吗 或者还有什么其他好的办法呢？