THIRDEYE: Cue-Aware Monocular Depth Estimation via Brain-Inspired Multi-Stage Fusion
This work addresses depth estimation for computer vision applications by leveraging explicit cues, but it appears incremental as it builds on existing cue-based approaches without presenting quantitative results.
The authors tackled monocular depth estimation by introducing ThirdEye, a pipeline that explicitly supplies monocular cues like occlusion boundaries and shading through pre-trained frozen networks, fused in a brain-inspired hierarchy with a working-memory module, resulting in a method that inherits external supervision and requires only modest fine-tuning.
Monocular depth estimation methods traditionally train deep models to infer depth directly from RGB pixels. This implicit learning often overlooks explicit monocular cues that the human visual system relies on, such as occlusion boundaries, shading, and perspective. Rather than expecting a network to discover these cues unaided, we present ThirdEye, a cue-aware pipeline that deliberately supplies each cue through specialised, pre-trained, and frozen networks. These cues are fused in a three-stage cortical hierarchy (V1->V2->V3) equipped with a key-value working-memory module that weights them by reliability. An adaptive-bins transformer head then produces a high-resolution disparity map. Because the cue experts are frozen, ThirdEye inherits large amounts of external supervision while requiring only modest fine-tuning. This extended version provides additional architectural detail, neuroscientific motivation, and an expanded experimental protocol; quantitative results will appear in a future revision.