Search for: All records

The state-of-the-art of fully-supervised methods for temporal action localization from untrimmed videos has achieved impressive results. Yet, it remains unsatisfactory for the weakly-supervised temporal action localization, where only video-level action labels are given without the timestamp annotation on when the actions occur. The main reason comes from that, the weakly-supervised networks only focus on the highly discriminative frames, but there are some ambiguous frames in both background and action classes. The ambiguous frames in background class are very similar to the real actions, which may be treated as target actions and result in false positives. On the other hand, the ambiguous frames in action class which possibly contain action instances, are prone to be false negatives by the weakly-supervised networks and result in a coarse localization. To solve these problems, we introduce a novel weakly-supervised Action Completeness Modeling with Back- ground Aware Networks (ACM-BANets). Our Background Aware Network (BANet) contains a weight-sharing two-branch architecture, with an action guided Background aware Temporal Attention Module (B-TAM) and an asymmetrical training strategy, to suppress both highly discriminative and ambiguous background frames to remove the false positives. Our action completeness modeling contains multiple BANets, and the BANets are forced to discover different but complementary action instances to completely localize the action instances in both highly discriminative and ambiguous action frames. In the 𝑖-th iteration, the 𝑖-th BANet discovers the discriminative features, which are then erased from the feature map. The partially-erased feature map is fed into the (i+1)-th BANet of the next iteration to force this BANet to discover discriminative features different from the 𝑖-th BANet. Evaluated on two challenging untrimmed video datasets, THUMOS14 and ActivityNet1.3, our approach outperforms all the current weakly-supervised methods for temporal action localization. 

Abstract Phytoplankton and associated microbial communities provide organic carbon to oceanic food webs and drive ecosystem dynamics. However, capturing those dynamics is challenging. Here, an in situ, semi-Lagrangian, robotic sampler profiled pelagic microbes at 4 h intervals over ~2.6 days in North Pacific high-nutrient, low-chlorophyll waters. We report on the community structure and transcriptional dynamics of microbes in an operationally large size class (>5 μm) predominantly populated by dinoflagellates, ciliates, haptophytes, pelagophytes, diatoms, cyanobacteria (chiefly Synechococcus), prasinophytes (chiefly Ostreococcus), fungi, archaea, and proteobacteria. Apart from fungi and archaea, all groups exhibited 24-h periodicity in some transcripts, but larger portions of the transcriptome oscillated in phototrophs. Periodic photosynthesis-related transcripts exhibited a temporal cascade across the morning hours, conserved across diverse phototrophic lineages. Pronounced silica:nitrate drawdown, a high flavodoxin to ferredoxin transcript ratio, and elevated expression of other Fe-stress markers indicated Fe-limitation. Fe-stress markers peaked during a photoperiodically adaptive time window that could modulate phytoplankton response to seasonal Fe-limitation. Remarkably, we observed viruses that infect the majority of abundant taxa, often with total transcriptional activity synchronized with putative hosts. Taken together, these data reveal a microbial plankton community that is shaped by recycled production and tightly controlled by Fe-limitation and viral activity.