The dynamics of music are one of the main elements that explains the characteristics of performances. They have been widely used for performance analysis, but they have not yet been thoroughly investigated in automatic music transcription. This paper proposes a system for estimating the intensity of individual notes from piano recordings. The algorithm is based on a score-informed non-negative matrix factorization (NMF) that takes the spectrogram of an audio recording and a corresponding MIDI score as inputs and factorizes the spectrogram into a set of spectral templates and their activations. The intensity of each note is obtained from the maximum activation of the corresponding pitch template around the onset of the note. We improved our previous system by employing an NMF model that can learn the temporal progress of timbre of piano notes. While our previous research was evaluated only with perfectly aligned scores, this paper also presents an evaluation with coarsely aligned scores. The result shows that our system is robust in aligning errors within 100 ms.