coco-minitrain
coco-minitrain copied to clipboard
giddyyupp
a subset of coco dataset for faster experimentation
Add download link [coco_minitrain_25k.zip](https://drive.google.com/file/d/1t_l9uyBPfxSEzcajTk4a1TaQXzeRm9hw/view?usp=sharing) https://github.com/giddyyupp/coco-minitrain/issues/15 https://github.com/giddyyupp/coco-minitrain/issues/18
One clarification question: is the link of [ instances_minitrain2017.json](https://drive.google.com/file/d/1lezhgY4M_Ag13w0dEzQ7x_zQ_w0ohjin/view) you shared is the 25k training samples that has been sampled by ```sample_coco.py``` for ```10M``` times? Then I don't need to...
i use sample_coco.py to sample my own mini-coco dataset for my experiment, it helps me save lots of time. Thanks for your work! Starred and wanted to Starred more than...
``` Traceback (most recent call last): File "sample_coco.py", line 156, in main() File "sample_coco.py", line 62, in main for i in keys[:parser.sample_image_count]: TypeError: slice indices must be integers or None...
Error when running ``` python3 coco_download.py --annotation train2017_mini_25k.json --output train2017_mini_25k ``` ``` Traceback (most recent call last): File "coco_download.py", line 23, in assert pathlib.Path(args.output_dir).is_dir(), "not valid dir" AssertionError: not valid...
I want to benchmark my model via coco-minitrain. As you know, training whole coco data is too slow to develop models. So I decided to use coco-minitrain. And you submitted...
由于网络限制，不能从coco官网抽样下载，所以 我更改了下代码，可以直接从 完整的train数据中， 根据作者提哦的instances_minitrain2017.json 将图片抽取出来 ``` import os from pycocotools.coco import COCO import wget import concurrent.futures import argparse import pathlib from shutil import copy parser = argparse.ArgumentParser(description="Download COCO images") parser.add_argument(...
hello there! I am currently attempting to train COCO minitrain on the CenterNet-better algorithm. However, as shown in the image below, I am required to have the person_keypoints_val_2017.json AND person_keypoints_train_2017.json...
Thank you for your excellent project! I'm training the keypoint RCNN network implemented in torchvision by scaling the image to the size denoted in object detection result tables. I know...
