Lameness in poultry is a serious economic and animal health issue. Real-time monitoring of the health of chickens using images is an efficient technique to stop widespread disease outbreaks due to the advancement in digital image processing and deep learning techniques. The spread of diseases among poultry usually poses a major threat to human health. Even if there are many people raising chickens in a lot of places, manual observation is still the main method used to monitor poultry diseases. The welfare of a broiler flock is often determined by factors such as mortality, physiology, behavior, and walking ability. Using a deep learning approach, it is possible to identify healthy and lame chickens. This study aims to design a convolutional neural network (CNN) model for identifying lame broilers within a poultry flock. Real-time data is obtained from two separate poultry farms, and data augmentation methods are applied to expand the dataset. The collected data is then manually categorized into two groups: healthy chickens and lame chicks. The model was developed using a dataset comprising 4,966 images of broilers, including 2,145 labeled as lame and 2,821 labeled as healthy. The goal is to categorize the birds based on their legs. The accuracy achieved for Convolutional Neural Networks model is 95%. The results demonstrate that lame broilers can be automatically detected within a flock, highlighting the potential of this method to improve flock management practices.