Building Flyweight FLIM-based CNNs with Adaptive Decoding for Object Detection

TL;DR

This paper introduces a lightweight, adaptive CNN built from user-drawn markers for object detection, achieving comparable or better performance than heavier models, suitable for resource-constrained environments.

Contribution

It extends FLIM to include an adaptive decoder that varies with input, enabling flyweight CNNs without backpropagation for object detection.

Findings

CNN weighs thousands of times less than SOTA detectors

Achieves superior or equivalent performance in five measures

Suitable for CPU execution in resource-constrained settings

Abstract

State-of-the-art (SOTA) object detection methods have succeeded in several applications at the price of relying on heavyweight neural networks, which makes them inefficient and inviable for many applications with computational resource constraints. This work presents a method to build a Convolutional Neural Network (CNN) layer by layer for object detection from user-drawn markers on discriminative regions of representative images. We address the detection of Schistosomiasis mansoni eggs in microscopy images of fecal samples, and the detection of ships in satellite images as application examples. We could create a flyweight CNN without backpropagation from very few input images. Our method explores a recent methodology, Feature Learning from Image Markers (FLIM), to build convolutional feature extractors (encoders) from marker pixels. We extend FLIM to include a single-layer adaptive decoder, whose weights vary with the input image -- a concept never explored in CNNs. Our CNN weighs thousands of times less than SOTA object detectors, being suitable for CPU execution and showing superior or equivalent performance to three methods in five measures.