Accuracy Comparison of Different Batch Size for a Supervised Machine Learning Task with Image Classification

SpiderID_APP: A User-Friendly APP for Spider Identification in Taiwan Using YOLO-Based Deep Learning Models

“…Input image size impacts computational requirements and detail detection, with smaller sizes reducing memory consumption and training time but potentially losing critical information. In comparison, larger sizes allow for finer detail detection but increase resource usage [76][77][78]. Furthermore, using an image size configuration higher than the training dataset image size can also cause potential loss of mid-level and high-level features [76], negatively affecting the model learning process and leading to suboptimal detection performance.…”

Teachable Machine: Optimization of Herbal Plant Image Classification Based on Epoch Value, Batch Size and Learning Rate

“…For example, users can collect images of different types of herbal plants and use Teachable Machine to train a model that can recognize and classify these plants based on images of these objects. This process can be optimized by adjusting parameters such as the number of epochs (iterations through the entire dataset), batch size (the number of samples processed before the model is updated), and learning rate (how fast the model learns) [36], [37], [38], [39], [40]. Furthermore, this model can be integrated into a mobile web application, allowing users to identify herbs by simply taking an image of the object.…”