深度学习目标检测模型学习报告：Faster-RCNN 和 YOLO 模型训练与调试

During the first stage of my learning process, I focused on acquiring the fundamental knowledge of deep learning neural networks that is required for my graduation project. To achieve this, I utilized various learning resources such as online courses and related literature. Additionally, I spent time researching open-source projects and reading relevant papers to gain a deeper understanding of the principles behind popular object detection models, such as Faster-RCNN and YOLO.

In terms of gathering data, I carefully selected 1000 images from the MSCOCO 2017 test dataset that had a suitable number of detection boxes. To ensure that the detection boxes were of appropriate size relative to the image, I resized each image to a size of 500 x 500 pixels and saved them in the .png format. I then utilized this curated dataset to train and fine-tune the Faster-RCNN and YOLO models.

Through this process, I was able to gain a comprehensive understanding of the theoretical concepts behind deep learning neural networks and object detection models, as well as practical experience working with real-world datasets and implementing these models in my project. I am excited to continue building upon this foundation in the next stage of my learning journey.