Yoga is a holistic practice that encompasses physical, mental, and spiritual well-being. Central to yoga are its postures, or “asanas,” which are designed to promote flexibility, strength, and balance in the body. Each yoga posture holds its own significance, targeting specific muscle groups and energy centers. Proper alignment in these postures is crucial not only for reaping physical benefits but also for cultivating mindfulness and concentration. In this work, we presented an approach to detect the yoga poses using CNN and LSTM methods. For each model, the data loading process involves reading images, resizing them, and organizing them into appropriate formats for training and testing. The CNN model comprises several convolutional and pooling layers, while the LSTM model includes LSTM layers for temporal sequence processing. The combined CNN-LSTM model emerged as a robust solution, leveraging the strengths of both architectures to achieve a comprehensive understanding of yoga poses. Its performance was evaluated on a test dataset, revealing insights into its accuracy and loss metrics. Notably, the CNN model exhibited an accuracy of 85.53%, surpassing other models studied in the research. This highlights the efficacy of the CNN approach in recognizing spatial features inherent in yoga postures, contributing significantly to the overall success of the model in identifying and categorizing yoga stances from input images.