SID-Net is a novel single image dehazing network that integrates adversarial and contrastive learning to improve dehazing performance. The network consists of three core branches: Image Dehazing Branch (IDB), Adversarial Guidance Branch (AGB), and Contrastive Enhancement Branch (CEB). IDB uses an encoder-decoder framework to perform haze-clean translation and enhances feature representation through the Attentive Recurrent Module (ARM) and Attention Fusion Operation (AFO). AGB leverages adversarial learning to guide the restored image toward a haze-free domain using clean ground truth. CEB employs contrastive learning to exploit negative information from hazy images, improving dehazing performance by pulling similar distributions closer and pushing dissimilar ones apart. Extensive experiments on synthetic and real-world datasets show that SID-Net achieves comparable results to state-of-the-art algorithms. The method addresses the limitation of existing learning-based approaches, which only use positive supervision from clean images and neglect negative information from hazy images. SID-Net effectively combines adversarial and contrastive learning to enhance dehazing performance. The network's design allows it to better utilize both positive and negative information, leading to improved results. The code is available at https://github.com/leandepk/SID-Net-for-image-dehazing. Keywords: Image dehazing, Convolutional neural networks, Adversarial learning, Contrastive learning.SID-Net is a novel single image dehazing network that integrates adversarial and contrastive learning to improve dehazing performance. The network consists of three core branches: Image Dehazing Branch (IDB), Adversarial Guidance Branch (AGB), and Contrastive Enhancement Branch (CEB). IDB uses an encoder-decoder framework to perform haze-clean translation and enhances feature representation through the Attentive Recurrent Module (ARM) and Attention Fusion Operation (AFO). AGB leverages adversarial learning to guide the restored image toward a haze-free domain using clean ground truth. CEB employs contrastive learning to exploit negative information from hazy images, improving dehazing performance by pulling similar distributions closer and pushing dissimilar ones apart. Extensive experiments on synthetic and real-world datasets show that SID-Net achieves comparable results to state-of-the-art algorithms. The method addresses the limitation of existing learning-based approaches, which only use positive supervision from clean images and neglect negative information from hazy images. SID-Net effectively combines adversarial and contrastive learning to enhance dehazing performance. The network's design allows it to better utilize both positive and negative information, leading to improved results. The code is available at https://github.com/leandepk/SID-Net-for-image-dehazing. Keywords: Image dehazing, Convolutional neural networks, Adversarial learning, Contrastive learning.