zhangst431 / TPRO

12 stars 1 forks source link

segment #2

Open geek-APTX4869 opened 6 months ago

geek-APTX4869 commented 6 months ago

Dear author: 似乎生成的伪标签对应的是test里的图像,而config里指定对train进行训练 def getitem(self, index): img_path = self.img_paths[index] mask_path = self.mask_paths[index] cls_label = self.cls_labels[index] assert os.path.exists(img_path), "image path: {}, does not exist".format(img_path) assert os.path.exists(mask_path), "mask path: {}, does not exist".format(mask_path)

报错,‘AssertionError: mask path: ./work_dirs/luad/classification/predictions\1032888-17499-16414-[1 0 0 1].png, does not exist’

zhangst431 commented 6 months ago

你好,在提取伪标签的命令中加一个参数:--split train,即可。README已经更新,感谢你提出的问题。

LUAD-HistoSeg
python evaluate_cls.py --dataset luad --model_path path/to/classification/model --save_dir ./work_dirs/luad/classification/predictions --split train
BCSS-WSSS
python evaluate_cls.py --dataset bcss --model_path path/to/classification/model --save_dir ./work_dirs/bcss/classification/predictions --split train
geek-APTX4869 commented 6 months ago

你好,在提取伪标签的命令中加一个参数:--split train,即可。README已经更新,感谢你提出的问题。

LUAD-HistoSeg
python evaluate_cls.py --dataset luad --model_path path/to/classification/model --save_dir ./work_dirs/luad/classification/predictions --split train
BCSS-WSSS
python evaluate_cls.py --dataset bcss --model_path path/to/classification/model --save_dir ./work_dirs/bcss/classification/predictions --split train

表1的结果是在哪个数据集划分下进行的呢?因为我发现加上'--split train'后,由于train划分没有mask,evaluate_cls.py 似乎不适用了,无法评估iou