Articles by tag: object-detection

TAPe detection: 100k parameters vs SOTA models

How the TAPe detector with 100k parameters outperforms YOLO and RF-DETR in speed and resource savings. Detailed breakdown of benchmarks and advantages for developers.

Bbox augmentation in Albumentations: formats and errors

Setting up A.BboxParams for detection: coordinate formats, label_fields, filtering. Avoid typical errors in the pipeline. Guide for developers.

YOLOE: object detection without fine-tuning

Learn how YOLOE uses text and visual prompts for detecting arbitrary objects. Backbone YOLO + RepRTA for speed. Code examples and scenarios for middle/senior dev. Test on your own tasks.

Superframe in YOLOv11: detection in video

Learn how superframe adds temporal context to YOLOv11 and RT-DETR for object detection in sports video. Experiments, code, results — for middle/senior dev.

TAPe without transformers: patch associations

Learn how abandoning transformers in TAPe+ML reduced parameters by orders of magnitude and gave the beginnings of segmentation on COCO. Local associations for object detection — read the details for developers.