<p>Although the field of underwater image enhancement has achieved notable progress, challenges such as blurred details, low color contrast, and insufficient brightness still hinder the advancement of current technologies. To address these issues, this paper proposes an underwater image enhancement network, termed DUAC-Net, which integrates a dual-attention Transformer with adaptive color correction. Specifically, a dual-attention synergistic Transformer block is designed and combined with a channel-wise color attention mechanism developed in this study to effectively extract and allocate critical features, thereby enhancing the model’s ability to represent color information. To further improve the representation of fine details, a multi-scale depthwise feature extraction module is introduced, enabling efficient multi-scale feature extraction by integrating multi-path information. To tackle the common problems of low contrast and insufficient brightness in underwater images, an improved adaptive color correction module is developed, which dynamically adjusts image contrast and brightness through learnable parameters. Furthermore, discrete wavelet transform is incorporated to separate the low- and high-frequency components of the image, thereby enriching the diversity and resolution of feature representations. Extensive experimental results demonstrate that DUAC-Net significantly outperforms existing state-of-the-art methods on three references and three non-reference datasets.</p>