00_2675.npy
stringlengths
3
11
clock
stringclasses
96 values
00_1583.npy
helmet
00_4759.npy
clock
00_0809.npy
earphone
00_4686.npy
bookshelf
00_1022.npy
helmet
00_4520.npy
clock
00_3985.npy
computer_keyboard
00_4405.npy
guitar
00_3685.npy
helmet
00_0187.npy
clock
00_5076.npy
helmet
00_1539.npy
clock
00_2829.npy
helmet
00_3073.npy
earphone
00_0985.npy
stove
00_0323.npy
clock
00_3749.npy
clock
00_5085.npy
clock
00_1798.npy
helmet
00_4834.npy
bus
00_3349.npy
earphone
00_4014.npy
clock
00_1472.npy
helmet
00_2575.npy
clock
00_4228.npy
earphone
00_4553.npy
helmet
00_5153.npy
clock
00_2279.npy
stove
00_2833.npy
helmet
00_5221.npy
loudspeaker
00_3246.npy
clock
00_2173.npy
helmet
00_0914.npy
stove
00_3591.npy
clock
00_0176.npy
helmet
00_3105.npy
earphone
00_1592.npy
washer
00_3681.npy
clock
00_4556.npy
helmet
00_2849.npy
clock
00_4966.npy
car
00_1896.npy
helmet
00_3062.npy
clock
00_3512.npy
tower
00_3274.npy
guitar
00_0274.npy
clock
00_2038.npy
clock
00_3849.npy
basket
00_2069.npy
helmet
00_2346.npy
basket
00_1479.npy
car
00_0587.npy
helmet
00_3882.npy
earphone
00_3025.npy
earphone
00_0436.npy
helmet
00_3608.npy
clock
00_4096.npy
earphone
00_1873.npy
loudspeaker
00_1446.npy
helmet
00_1098.npy
car
00_4323.npy
earphone
00_4261.npy
clock
00_2738.npy
cellular_telephone
00_2622.npy
car
00_4956.npy
clock
00_4767.npy
clock
00_2315.npy
car
00_3870.npy
stove
00_4813.npy
remote_control
00_1569.npy
clock
00_0814.npy
clock
00_4636.npy
bus
00_3201.npy
helmet
00_4223.npy
helmet
00_4825.npy
helmet
00_4007.npy
stove
00_2142.npy
clock
00_2753.npy
clock
00_2577.npy
car
00_1860.npy
earphone
00_4454.npy
clock
00_1960.npy
washer
00_5165.npy
loudspeaker
00_0318.npy
clock
00_1374.npy
stove
00_3494.npy
clock
00_3296.npy
clock
00_2935.npy
clock
00_1169.npy
earphone
00_2975.npy
clock
00_0503.npy
earphone
00_1919.npy
car
00_1542.npy
clock
00_1596.npy
clock
00_0129.npy
camera
00_0492.npy
laptop
00_2139.npy
helmet
00_0348.npy
clock
00_2265.npy
loudspeaker
00_0782.npy
clock
YAML Metadata Warning: empty or missing yaml metadata in repo card (https://huggingface.co/docs/hub/datasets-cards)

数据集说明

数据集结构

├ trainset/
├── 0/
│   ├── [label]-xxxxx.npy
│   ├── ......
├── 1/
├── ......
├── 10/
│   
├ testset/
├── 0/
│   ├── xxxxx.npy
│   ├── ......
├── 1/
├── ......
├── 10/

文件夹"0"为基类数据集,文件夹"1"-"10"为新类数据集的10个增量阶段数据。每个文件夹内包含.npy格式的点云数据,可以使用numpy.load(path)读取相应文件。在trainset中,文件命名保证依照"[label]-xxxxx.npy"格式,其中第一个"-"前的内容为该点云的类别名,可以使用filename.split('-')[0]获取相应标签。

数据集读取示例

class CONTESTCIL_train(Dataset):
    def __init__(self, root='./trainset', session=0, name2id=None):
        data_root = os.path.join(root,str(session),"train")
        self.point_cloud = []
        self.labels = []
        self.session = session
        self.session_label_num = [55,59,63,67,71,75,79,83,87,91,95]
        for name in os.listdir(self.data_root):
            self.point_cloud.append(os.path.join(data_root,name))
            self.labels.append(name2id[name.split('-')[0]])
        
    def get_cat_num(self):
        return self.session_label_num[self.session]
    
    def __getitem__(self, idx):
        return np.load(self.point_cloud[idx]), self.labels[idx]

    def __len__(self):
        return len(self.labels)


class CONTESTCIL_test(Dataset):
    def __init__(self, root='./testset', session=0):
        self.session = session
        self.session_label_num = [55,59,63,67,71,75,79,83,87,91,95]
        data_root = os.path.join(root,str(session))
        self.point_cloud = []
        self.name = []
        for name in os.listdir(self.data_root):
            self.point_cloud.append(os.path.join(data_root,name))
            self.name.append(name)
    
    def get_cat_num(self):
        return self.session_label_num[self.session] 
    
    def __getitem__(self, idx):
        return np.load(self.point_cloud[idx]),self.name[idx]

    def __len__(self):
        return len(self.point_cloud)

答案提交

参赛选手需要提交 test.csv,其所需内容要求如下:

  • 选手提交的答案文件,每行均包含两列,其中第一列为文件名称"XXX.npy"(输出内容无需加双引号),第二列为选手模型预测的标签,如"sofa"等。
  • 选手对于每个增量阶段,均需要对该阶段测试集所有样本进行预测并输出。在每个增量阶段所有样本答案输出结束之后,请输出一行"END,END"用于标记该增量阶段结束。请注意:无效文件名,相同文件名重复输出均不会被统计。
  • 具体格式可以参考主办方发布的"test.csv"文件,若存在提交格式问题需要确认,请联系主办方。
Downloads last month
5
Edit dataset card