Fatigue is a neuromuscular condition wherein the muscle fails to produce the required force. Analysis of this condition is crucial in the field of myoelectric control, sports medicine and ergonomics. Muscle biopsy is widely used for the assessment of muscle fatigue in clinical diagnosis. However, this technique may not be suitable for other applications because of its invasiveness. Surface Electromyography (sEMG) is a noninvasive technique that records the electrical activity of muscle contractions. These signals reflect the underlying physiological state of the muscle dynamics. But, the extraction of reliable information from these signals are considered to be challenging as these signals exhibit the nonlinear and nonstationary behaviour. This research aims at enhancing the applicability of sEMG signals in the detection of muscle fatigue conditions using shape descriptors. The objectives of this research work are to extract the geometric features and to develop models for the detection of muscle fatigue conditions in isometric and dynamic contractions. The shape of the signals is computed using geometric features that are extracted from four domains namely time, frequency, time-frequency and frequency-frequency domains. The information on shape is explored to develop machine learning detection models for the differentiation of nonfatigue and fatigue conditions. In this seminar, key results are extracted from all the four domains for both the contractions and their classification performances are discussed in detail.