load_i2e_datasets.py

import io
import torch
import numpy as np
from datasets import load_dataset
from torch.utils.data import Dataset


# =================================================
# Read and unpack function (Your proposed Refactor)
# =================================================
def unpack_event_data(item, use_io=True):
    if use_io:
        with io.BytesIO(item['data']) as f:
            raw_data = np.load(f)
    else:
        raw_data = np.load(item)
    header_size = 4 * 2                                         # Parse Header (First 4*2=8 bytes)
    shape_header = raw_data[:header_size].view(np.uint16)       # Assume shape is 4D (T, C, H, W)
    original_shape = tuple(shape_header)
    
    packed_body = raw_data[header_size:]                        # Parse Body
    unpacked = np.unpackbits(packed_body)                       # Unpack bits to binary array
    
    num_elements = np.prod(original_shape)                      # Extract valid bits (Handle padding)
    event_flat = unpacked[:num_elements]
    
    event_data = event_flat.reshape(original_shape).astype(np.float32).copy()   # Reshape and convert to Float Tensor
    return torch.from_numpy(event_data)


class I2E_Dataset(Dataset):
    def __init__(self, cache_dir, load_datasets, split='train', transform=None, target_transform=None):
        self.ds = load_dataset('UESTC-BICS/I2E', load_datasets, split=split, cache_dir=cache_dir)
        self.transform = transform
        self.target_transform = target_transform

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

    def __getitem__(self, idx):
        item = self.ds[idx]
        event = unpack_event_data(item)      # Returns [T, C, H, W] Tensor
        label = item['label']

        if self.transform:
            event = self.transform(event)
        if self.target_transform:
            label = self.target_transform(label)
        
        return event, label