alen-smajic / Real-time-Object-Detection-for-Autonomous-Driving-using-Deep-Learning

My Computer Vision project from my Computer Vision Course (Fall 2020) at Goethe University Frankfurt, Germany. Performance comparison between state-of-the-art Object Detection algorithms YOLO and Faster R-CNN based on the Berkeley DeepDrive (BDD100K) Dataset.

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Low performances of YOLO (mAP)

Sarah-Leroy opened this issue · comments

Hello,

First, I wanted to thank you for sharing your work, which I am using in my internship.

For FasterRCNN I find a similar mAP but not for YOLO (2% vs 18%). I checked all the different steps and I can't find the problem...
I attached some detections files (format: class confidence x1 y1 x2 y2 absolute), output images (blue: ground truth correctly detected, green: correct detection, red: false detection, pink: ground truth not detected) and maP results for 1000 images.
Do you have any ideas to help me?

Thank you.

b1db7e22-cfa74dc3.txt
b1dce572-c6a8cb5e.txt
b1cebfb7-284f5117.txt
b1d7b3ac-9e14f05f.txt
b1d0091f-f2c2d2ae.txt
b1db7e22-cfa74dc3
b1dce572-c6a8cb5e
b1cebfb7-284f5117
b1d7b3ac-9e14f05f
b1d0091f-f2c2d2ae
output.txt