Estimating the preferences of consumers is of utmost importance for the fashion industry as appropriately leveraging this information can be beneficial in terms of profit. Trend detection in fashion is a challenging task due to the fast pace of change in the fashion industry. Moreover, forecasting the visual popularity of new garment designs is even more demanding due to lack of historical data. To this end, we propose MuQAR, a Multimodal Quasi-AutoRegressive deep learning architecture that combines two modules: (1) a multi-modal multi-layer perceptron processing categorical and visual features extracted by computer vision networks and (2) a quasi-autoregressive neural network modelling the time series of the product's attributes, which are used as a proxy of temporal popularity patterns mitigating the lack of historical data. We perform an extensive ablation analysis on two large scale image fashion datasets, Mallzee-popularity and SHIFT15m to assess the adequacy of MuQAR and also use the Amazon Reviews: Home and Kitchen dataset to assess generalisability to other domains. A comparative study on the VISUELLE dataset, shows that MuQAR is capable of competing and surpassing the domain's current state of the art by 2.88% in terms of WAPE and 3.04% in terms of MAE.