Irene Amerini
Papers
4
Total Citations
30
H-Index
3
About
Irene Amerini is a leading researcher in computer vision and embedded AI, specializing in monocular depth estimation (MDE) — the task of inferring depth from a single RGB image. Her work addresses a critical bottleneck in autonomous systems, robotics, and IoT: enabling real-time, energy-efficient depth perception on resource-constrained devices. Amerini’s major contributions include the development of SPEED, a Separable Pyramidal Pooling Encoder-Decoder architecture that achieves real-time MDE on low-resource platforms, and pioneering lightweight, energy-aware models for both terrestrial and underwater environments. Her research systematically optimizes Vision Transformer architectures with efficient attention modules, balancing accuracy against power consumption. With over 30 citations across her most-cited papers (including 16 for SPEED), her impact is growing rapidly. Notably, she has advanced the deployment of deep learning on IoT and embedded devices, demonstrating that state-of-the-art depth estimation is feasible in underwater scenarios — a notoriously challenging domain. Amerini’s work is essential for students and engineers building autonomous robots, drones, or underwater vehicles that must “see” in 3D without heavy computational overhead.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3
- 4