This paper introduces a novel deep hashing network for unsupervised domain adaptation, called Domain Adaptive Hashing (DAH). The DAH framework leverages the feature learning capabilities of deep neural networks to learn efficient hash codes for domain adaptation tasks. The proposed method addresses the challenge of transferring knowledge from a source domain with labeled data to a target domain with unlabeled data. The DAH framework incorporates three key components: (i) a supervised hash loss for labeled source data, (ii) an unsupervised entropy loss for unlabeled target data, and (iii) a multi-kernel Maximum Mean Discrepancy (MK-MMD) loss to minimize the distribution difference between the source and target domains. The DAH network is trained using a combination of these losses to learn discriminative hash codes that enable accurate classification of the target domain. The paper also introduces a new dataset, Office-Home, which contains images from four domains and is used to evaluate the DAH algorithm. The dataset includes around 15,500 images organized into 65 categories. The DAH algorithm is evaluated on multiple transfer tasks, and the results show that it outperforms existing competitive baselines for unsupervised domain adaptation. The DAH framework is able to learn efficient hash codes that are representative of the source and target domains, leading to improved classification accuracy. The DAH framework is implemented as a deep convolutional neural network with five convolutional layers and three fully connected layers. The network is trained using a combination of supervised and unsupervised losses to learn hash codes that are discriminative and robust. The hash codes are generated using a tanh activation function to ensure they are bounded between -1 and +1. The DAH framework is also evaluated on the Office dataset, and the results show that it outperforms other state-of-the-art domain adaptation methods, including Geodesic Flow Kernel (GFK), Transfer Component Analysis (TCA), Correlation Alignment (CORAL), and Joint Distribution Adaptation (JDA). The DAH framework is also compared with deep learning methods such as Deep Adaptation Network (DAN) and Domain Adversarial Neural Network (DANN), and it is shown to perform better in terms of classification accuracy. The DAH framework is also evaluated for unsupervised domain adaptive hashing, where the goal is to generate compact and efficient hash codes for classifying unseen test instances without any labels. The results show that the DAH framework outperforms other unsupervised hashing methods, including ITQ, KMeans, BA, and BDNN. The DAH framework is able to leverage the labeled data in the source domain to learn efficient hash codes for the target domain, leading to improved performance. The proposed DAH framework is able to learn representative hash codes by utilizing labeled data from a different domain, making it a valuable tool for real-world applications.This paper introduces a novel deep hashing network for unsupervised domain adaptation, called Domain Adaptive Hashing (DAH). The DAH framework leverages the feature learning capabilities of deep neural networks to learn efficient hash codes for domain adaptation tasks. The proposed method addresses the challenge of transferring knowledge from a source domain with labeled data to a target domain with unlabeled data. The DAH framework incorporates three key components: (i) a supervised hash loss for labeled source data, (ii) an unsupervised entropy loss for unlabeled target data, and (iii) a multi-kernel Maximum Mean Discrepancy (MK-MMD) loss to minimize the distribution difference between the source and target domains. The DAH network is trained using a combination of these losses to learn discriminative hash codes that enable accurate classification of the target domain. The paper also introduces a new dataset, Office-Home, which contains images from four domains and is used to evaluate the DAH algorithm. The dataset includes around 15,500 images organized into 65 categories. The DAH algorithm is evaluated on multiple transfer tasks, and the results show that it outperforms existing competitive baselines for unsupervised domain adaptation. The DAH framework is able to learn efficient hash codes that are representative of the source and target domains, leading to improved classification accuracy. The DAH framework is implemented as a deep convolutional neural network with five convolutional layers and three fully connected layers. The network is trained using a combination of supervised and unsupervised losses to learn hash codes that are discriminative and robust. The hash codes are generated using a tanh activation function to ensure they are bounded between -1 and +1. The DAH framework is also evaluated on the Office dataset, and the results show that it outperforms other state-of-the-art domain adaptation methods, including Geodesic Flow Kernel (GFK), Transfer Component Analysis (TCA), Correlation Alignment (CORAL), and Joint Distribution Adaptation (JDA). The DAH framework is also compared with deep learning methods such as Deep Adaptation Network (DAN) and Domain Adversarial Neural Network (DANN), and it is shown to perform better in terms of classification accuracy. The DAH framework is also evaluated for unsupervised domain adaptive hashing, where the goal is to generate compact and efficient hash codes for classifying unseen test instances without any labels. The results show that the DAH framework outperforms other unsupervised hashing methods, including ITQ, KMeans, BA, and BDNN. The DAH framework is able to leverage the labeled data in the source domain to learn efficient hash codes for the target domain, leading to improved performance. The proposed DAH framework is able to learn representative hash codes by utilizing labeled data from a different domain, making it a valuable tool for real-world applications.