Fine-Tuning Deep Learning Models for Pedestrian Detection

Abstract

Object detection in high resolution images is a new challenge that the remote sensing community is facing thanks to introduction of unmanned aerial vehicles and monitoring cameras. One of the interests is to detect and trace persons in the images. Different from general objects, pedestrians can have different poses and are undergoing constant morphological changes while moving, this task needs an intelligent solution. Fine-tuning has woken up great interest among researchers due to its relevance for retraining convolutional networks for many and interesting applications. For object classification, detection, and segmentation fine-tuned models have shown state-of-the-art performance. In the present work, we evaluate the performance of fine-tuned models with a variation of training data by comparing Faster Region-based Convolutional Neural Network (Faster R-CNN) Inception v2, Single Shot MultiBox Detector (SSD) Inception v2, and SSD Mobilenet v2. To achieve the goal, the effect of varying training data on performance metrics such as accuracy, precision, F1-score, and recall are taken into account. After testing the detectors, it was identified that the precision and recall are more sensitive on the variation of the amount of training data. Under five variation of the amount of training data, we observe that the proportion of 60%-80% consistently achieve highly comparable performance, whereas in all variation of training data Faster R-CNN Inception v2 outperforms SSD Inception v2 and SSD Mobilenet v2 in evaluated metrics, but the SSD converges relatively quickly during the training phase. Overall, partitioning 80% of total data for fine-tuning trained models produces efficient detectors even with only 700 data samples.

Custom Object Detection Using Transfer Learning with Pretrained Models for Improved Detection Techniques

“…The fine-tuning approach is a commonly used technique in the field of custom object detection using transfer learning with pretrained models (Amisse et al, 2021). It involves adapting a pretrained model, which has been previously trained on a large-scale dataset, to perform custom object detection on a target dataset with limited labeled examples.…”

A Data-Driven Model for Pedestrian Behavior Classification and Trajectory Prediction

“…In the first approach, the training period is defined as a fixed percentage of the available dataset size, for example 50% of the observations are used for training and the rest for testing [35]. In the second approach, the optimal training period is determined by the desired prediction accuracy [36]. When the algorithm returns a low prediction error (lower than a set threshold values), the training process is accomplished at a satisfactory level.…”

Selective Layer Tuning and Performance Study of Pre-Trained Models Using Genetic Algorithm

“…In a heart sound classification study, pre-trained audio neural networks were fine-tuned based on a heart sound database [16]. Moreover, in a pedestrian detection study, using a mobile phone with a 4-MP camera, Faster R-CNN + InceptionV2, SSD + InceptionV2, and SSD + MobileNetV2 were fine-tuned based on street video data [17]. In addition, cutterhead torque prediction [18] and atrial fibrillation detection [19] automatically extract informative deep features from the source and target domains and then feed them into the feature predictor for final results.…”