Automated recognition of mouse behaviours is crucial in studying psychiatric and neurologic diseases. To achieve this objective, it is very important to analyse temporal dynamics of mouse behaviours. In particular, the change between mouse neighbouring actions is swift in a short period. In this paper, we develop and implement a novel Hidden Markov Model (HMM) algorithm to describe the temporal characteristics of mouse behaviours. In particular, we here propose a hybrid deep learning architecture, where the first unsupervised layer relies on an advanced spatial-temporal segment Fisher Vector (SFV) encoding both visual and contextual features. Subsequent supervised layers based on our segment aggregate network (SAN) are trained to estimate the state dependent observation probabilities of the HMM. The proposed architecture shows the ability to discriminate between visually similar behaviours and results in high recognition rates with the strength of processing imbalanced mouse behaviour datasets. Finally, we evaluate our approach using JHuang’s and our own datasets, and the results show that our method outperforms other state-of-the-art approaches.

Bottom-up, end-user based feed, and food analysis through smartphone quantification of lateral flow assays (LFA) has the potential to cause a paradigm shift in testing capabilities. However, most developed devices do not test the presence of and implications of inter-phone variation. Much discussion remains regarding optimum color space for smartphone colorimetric analyses and, an in-depth comparison of color space performance is missing. Moreover, a light-shielding box is often used to avoid variations caused by background illumination while the use of such a bulky add-on may be avoidable through image background correction. Here, quantification performance of individual channels of RGB, HSV, and LAB color space and ΔRGB was determined for color and color intensity variation using pH strips, filter paper with dropped nanoparticles, and colored solutions. LAB and HSV color space channels never outperformed the best RGB channels in any test. Background correction avoided measurement variation if no direct sunlight was used and functioned more efficiently outside a light-shielding box (prediction errors < 5%/35% for color/color intensity change). The system was validated using various phones for quantification of major allergens (i.e., gluten in buffer, bovine milk in goat milk and goat cheese), and, pH in soil extracts with commercial pH strips and LFA. Inter-phone variation was significant for LFA quantification but low using pH strips (prediction errors < 10% for all six phones compared). Thus, assays based on color change hold the strongest promise for end-user adapted smartphone diagnostics.