Filtering Empty Camera Trap Images in Embedded Systems

TL;DR

This paper compares animal recognition models for embedded wildlife camera traps, analyzing their accuracy and speed trade-offs, and proposes optimized models to efficiently filter empty images on resource-limited devices.

Contribution

It provides a comprehensive comparison of classifiers and detectors for empty image filtering, including optimization techniques and performance analysis on edge devices.

Findings

Detectors outperform classifiers in empty image removal with similar latency.

Classifiers can achieve comparable results to detectors with ten times more training data.

Optimized models using quantization and filter reduction improve inference speed and accuracy.

Abstract

Monitoring wildlife through camera traps produces a massive amount of images, whose a significant portion does not contain animals, being later discarded. Embedding deep learning models to identify animals and filter these images directly in those devices brings advantages such as savings in the storage and transmission of data, usually resource-constrained in this type of equipment. In this work, we present a comparative study on animal recognition models to analyze the trade-off between precision and inference latency on edge devices. To accomplish this objective, we investigate classifiers and object detectors of various input resolutions and optimize them using quantization and reducing the number of model filters. The confidence threshold of each model was adjusted to obtain 96% recall for the nonempty class, since instances from the empty class are expected to be discarded. The experiments show that, when using the same set of images for training, detectors achieve superior performance, eliminating at least 10% more empty images than classifiers with comparable latencies. Considering the high cost of generating labels for the detection problem, when there is a massive number of images labeled for classification (about one million instances, ten times more than those available for detection), classifiers are able to reach results comparable to detectors but with half latency.