The asymptotically optimal transmission rate of binary codes correcting localized errors is known when the number of errors grows linearly in the code length. Here we prove that this rate can be attained by codes with polynomial complexity of encoding, decoding, and code construction.