Efficient Majority Voting in Digital Hardware

Symmetric Multi-Scale Residual Network Ensemble with Weighted Evidence Fusion Strategy for Facial Expression Recognition

“…Decision fusion strategies mainly include majority vote and D-S evidence theory. The majority voting strategy is a simple and effective method for decision-level data fusion [40]. However, the majority voting method does not consider the importance and confidence of each individual, and is prone to controversial decision results.…”

Design of low-power gesture recognition system and self-organizing map processors

“…The majority voting unit (MVU) collects all the recognition results from the ECNNC (RE) and the DTC (RD) throughout a motion and selects the candidates whose votes dominate the collection of the result. The structure of the MVU is similar to [39]. In the first stage, two 3-bit input class labels from RC (RE & RD) are decoded into the one- hot codes, i.e., the 6-bit vectors including all '0's but a '1' at the position representing the gesture class.…”

“…Then two one-hot codes are split into six 2-bit pairs for controlling the counter of each class. Note that the parallel adder trees in [39] for calculating the total votes of each class are removed here. This is because only two classifiers are adopted for the classification of each frame instead of ensembling T (T > 10) decision trees as in [39], [40].…”

“…Note that the parallel adder trees in [39] for calculating the total votes of each class are removed here. This is because only two classifiers are adopted for the classification of each frame instead of ensembling T (T > 10) decision trees as in [39], [40]. Besides, as the collection for voting is based on multiple frames rather than one frame, the adder can be used in the time-multiplexing style.…”