Spaces:
Runtime error
Runtime error
File size: 1,761 Bytes
f15a1cd |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 |
from collections.abc import Sequence
from abc import ABC, abstractmethod
import torch
from PIL.Image import Image
class FeatureExtractor(ABC):
@abstractmethod
def encode_image(self, img_list: Sequence[Image]) -> torch.Tensor:
"""
Encode the input images and return the corresponding embeddings.
Args:
img_list: A list of PIL.Image.Image objects.
Returns:
The embeddings of the input images. The shape should be (len(img_list), embedding_dim).
"""
raise NotImplementedError
@abstractmethod
def encode_text(self, text_list: Sequence[str]) -> torch.Tensor:
"""
Encode the input text data and return the corresponding embeddings.
Args:
text_list: A list of strings.
Returns:
The embeddings of the input text data. The shape should be (len(text_list), embedding_dim).
"""
raise NotImplementedError
@abstractmethod
def encode_3D(self, pc_tensor: torch.Tensor) -> torch.Tensor:
"""
Encode the input 3D point cloud and return the corresponding embeddings.
Args:
pc_tensor: A tensor of shape (B, N, 3 + 3).
Returns:
The embeddings of the input 3D point cloud. The shape should be (B, embedding_dim).
"""
raise NotImplementedError
@abstractmethod
def encode_query(self, queries: Sequence[str]) -> torch.Tensor:
"""Encode the queries and return the corresponding embeddings.
Args:
queries: A list of strings.
Returns:
The embeddings of the input text data. The shape should be (len(input_text), embedding_dim).
"""
raise NotImplementedError |