search

PaddlePaddle / PaddleDetection

Object Detection toolkit based on PaddlePaddle. It supports object detection, instance segmentation, multiple object tracking and real-time multi-person keypoint detection.
Apache License 2.0
12.82k stars 2.89k forks source link

Is the models in modelZOO exported model? #1778

Closed athulvingt closed 1 year ago

athulvingt commented 3 years ago

DO I have to export the models in modelzoo for deployment purpose or use it as it is

yghstill commented 3 years ago

If you are only for testing purposes, just use it directly. And if you want to deploy, you should export model firstly and then use deploy/python/infer.py or deploy/cpp to deploy.

athulvingt commented 3 years ago

I was following instruction on QUICK_STARTED_cn.md python tools/infer.py -c configs/ppyolo/ppyolo.yml -o use_gpu=true weights=https://paddlemodels.bj.bcebos.com/object_detection/ppyolo.pdparams --infer_img=demo/000000014439.jpg

I downloaded ppyolo mode (weights and configuration file)l from model zoo and run

python tools/infer.py -c ppyolo/ppyolo.yml -o use_gpu=true weights=ppyolo/ppyolopdarams --infer_img=demo/000000014439.jpg

but I got file not found error - no file named ppyolo_reader.yml

yghstill commented 3 years ago

This is caused by the lack of ppyolo_reader.yml. Please copy configs/ppyolo/ppyolo_reader.yml to ppyolo folder.

athulvingt commented 3 years ago

I downloaded PP-yolo_2x ResNet50vd ,input size 320 from ppyolo model zoo and exported it. While inferencing the model on a video it used only 1 thread of my 16 thread CPU, but the inference time was too low (0.88 FPS). I would like to have a better inference time. Am I doing something wrong?

thinkthinking commented 1 year ago

You'd better use GPU to inference. Welcome to use PaddleDetection v2.6!