Object detection in fast.ai v1

tester · February 12, 2019, 6:38am

Seems like yolo is more famous. Just for curiosity, why would we pick retina net over yolov3 for lecture? Is retina net easier to teach or other reasons?

bjack913 · February 25, 2019, 6:19am

RetinaNet uses a pretrained ResNet backbone, while Yolov3 uses Darknet53 as a backbone. There is no pretrained Darknet53 model available in fastai, which you would need to implement Yolov3. This is why fastai will use RetinaNet.

FWIW, I tried to load the weights from the original Darknet53 implementation in C into the fastai implementation. I could never get it to quite work. I think the weights were loaded correctly, but since fastai uses a slightly different output layer than the original C implementation, I couldn’t validate it. I’ve since moved on to trying to train fastai Darknet53 on imagenet from scratch. I’ll share my results on the forum when I get the top-1 accuracy into the 70%+ range.

monocongo · March 8, 2019, 6:18pm

Thanks for this, nice work!

yangdf · March 10, 2019, 10:32am

I have a problem. When I run the code to transfer the pre-training model: learn=create_cnn(…), the code will automatically download the model, but it will be interrupted due to time and network speed, so I download the pre-training model in advance. When you run this code again, it will be stuck here. How can I solve this problem? Please help everyone.

yangdf · March 10, 2019, 10:35am

andresa · March 20, 2019, 12:34am

Hi Sgugger,
I am trying to run the dev notebook 102a_coco and I receive the error objectdetectdataset not defined how do I fix this error?
Thanks

charming · March 22, 2019, 7:49am

github.com

fastai/fastai_docs/blob/master/dev_course/dl2/pascal.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from fastai.vision import *"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Data"
   ]
  },
  {
   "cell_type": "markdown",

This file has been truncated. show original

there are new notebook for retinaNet Objection

Bronzi88 · April 14, 2019, 11:57am

Hi,

I tried to put all the pieces for object detection in fast.ai v1 with RetinaNet together.

With kind regrads,
Christian

vha14 · April 14, 2019, 3:18pm

Nice work! Have you tried to train on Pascal VOC and/or COCO datasets and measure mAPs?

Bronzi88 · April 14, 2019, 3:29pm

I put everything together to do so including writing a callback for fast.ai, but I´m a little bit short on computational resources to do so.

DollofCuty · April 28, 2019, 2:27pm

Thank you very much, this looks great!

My only problem are some imports in:

ObjectDetection/callbacks/callbacks.py

from BoundingBox import BoundingBox
from BoundingBoxes import BoundingBoxes
from Evaluator import *
from utils import *

I’ve never seen those BoundingBox classes before which leads me to think you wrote them yourself and did not commit them.
Also I can’t find Evaluator or utils in the repo.

vha14 · April 28, 2019, 7:59pm

You can find them from https://github.com/rafaelpadilla/Object-Detection-Metrics

Bronzi88 · April 29, 2019, 7:53am

Hey,

sorry for not pointing that out more clearly.
You have to install Object-Detection-Metrics as I mentioned in the readme.

With kind regards,
Christian

DollofCuty · April 30, 2019, 2:25pm

Oh no, my bad sorry, I overread that 3 times apparently.
Thank you for the clarification!

mohamed1 · May 15, 2019, 3:41pm

i tried running that notebook (pascal.ipynb )and only got 17% mAP. did anyone get better results?

champs.jaideep · May 25, 2019, 6:04am

hi kristen
I get an error in line 2 items self in original coco notebook
train_ds = ObjectDetectDataset.from_json(PATH/‘train2017’, ANNOT_PATH/‘train_sample.json’)

objectdetect Dataset etc are not available where are they defined ?
what is the purpose of bb_pad_collate used in
If we are already given the xy,xmax,ymax then we can directly use them without having read i from any json source ?

dhoa · May 28, 2019, 2:15pm

Very nice work @Bronzi88 . Thank you so much.

I have just tried your notebook and everything works great. Can I ask you one question ? What is the output of your Retina Net ? How can I get the bounding box from the output. I have just tried it with an image and the output is a list with 3 items. The 1st one size is [1, 24480, 3] so I guess it is related to the probability of the class, 2nd one size is [1, 24480, 4] so I think it is the bounding box. The last one is [[32, 32], [16, 16], [8, 8], [4, 4]]], I don’t know what it is.

I want to understand the output because I don’t use fast.ai for inference. Thank you so much

dhoa · May 29, 2019, 8:46am

I have digged in your code and understand how it works now. Thank you so much.

champs.jaideep · May 29, 2019, 10:36am

hi there…
Thanks for posting…
how did you install.obd metric .is there .git for installation

dhoa · May 29, 2019, 10:52am

You can just clone it and add your sys with Object-Detection-Metrics/lib/ . That’s what I did