Long-tail distribution is widely spread in real-world applications. Due to the extremely small ratio of instances, tail categories often show inferior accuracy. In this paper, we find such performance bottleneck is mainly caused by the imbalanced gradients, which can be categorized into two parts: (1) positive part, deriving from the samples of the same category, and (2) negative part, contributed by other categories. Based on comprehensive experiments, it is also observed that the gradient ratio of accumulated positives to negatives is a good indicator to measure how balanced a category is trained. Inspired by this, we come up with a gradient-driven training mechanism to tackle the long-tail problem: re-balancing the positive/negative gradients dynamically according to current accumulative gradients, with a unified goal of achieving balance gradient ratios. Taking advantage of the simple and flexible gradient mechanism, we introduce a new family of gradient-driven loss functions, namely equalization losses. We conduct extensive experiments on a wide spectrum of visual tasks, including two-stage/single-stage long-tailed object detection (LVIS), long-tailed image classification (ImageNet-LT, Places-LT, iNaturalist), and long-tailed semantic segmentation (ADE20K). Our method consistently outperforms the baseline models, demonstrating the effectiveness and generalization ability of the proposed equalization losses.Codes will be released at https://github.com/ModelTC/United-Perception.

No abstract

Animals behave differently in response to visual cues with distinct ethological meaning, a process usually thought to be achieved through differential visual processing. Using a defined zebrafish escape circuit as a model, we found that behavior selection can be implemented at the visuomotor transformation stage through a visually responsive dopaminergic-inhibitory circuit module. In response to non-threatening visual stimuli, hypothalamic dopaminergic neurons and their positively regulated hindbrain inhibitory interneurons increase activity, suppressing synaptic transmission from the visual center to the escape circuit. By contrast, threatening visual stimuli inactivate some of these neurons, resulting in dis-inhibition of the visuomotor transformation and escape generation. The distinct patterns of dopaminergic-inhibitory neural module's visual responses account for this stimulus-specific visuomotor transformation and behavioral control. Thus, our study identifies a behavioral relevance-dependent mechanism that controls visuomotor transformation and behavior selection and reveals that neuromodulation can be tuned by visual cues to help animals generate appropriate responses.