fiveai/hardest
The HARDness ESTimation package: A library for ranking images from a dataset by hardness with respect to a specific detector.
This tool helps machine learning engineers and researchers quickly identify the most challenging images for an object detection model to accurately process. You provide a dataset of images and the model's predictions for those images. The output is a ranked list of images, from 'hardest' to 'easiest', showing where your model struggles most.
No commits in the last 6 months.
Use this if you need to understand which types of images or scenarios your object detection model performs poorly on, or to select difficult examples for further model training or evaluation.
Not ideal if you need to improve the model's accuracy directly; this tool helps identify problematic data but doesn't fix the model itself.
Stars
18
Forks
1
Language
Python
License
—
Category
Last pushed
Jan 03, 2023
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/fiveai/hardest"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
DeepTrackAI/DeepTrack2
DeepTrack2 is a modular Python library for generating, manipulating, and analyzing image data...
abhineet123/Deep-Learning-for-Tracking-and-Detection
Collection of papers, datasets, code and other resources for object tracking and detection using...
NVIDIA-ISAAC-ROS/isaac_ros_dnn_inference
NVIDIA-accelerated DNN model inference ROS 2 packages using NVIDIA Triton/TensorRT for both...
DagnyT/hardnet
Hardnet descriptor model - "Working hard to know your neighbor's margins: Local descriptor...
rafellerc/Pytorch-SiamFC
Pytorch implementation of "Fully-Convolutional Siamese Networks for Object Tracking"