search

sowmyadvn / Crime-Detection--using-Deep-learning

Developed an object detection model using YOLO Darknet framework to prevent crimes. Collected the crime data manually from Google images as well as Imagenet database
Other
51 stars 31 forks source link

.weights #3

Open evertonesm opened 5 years ago

evertonesm commented 5 years ago

Sorry but is the weigths files available to test the detection?

moeiscool commented 5 years ago

i am also looking for it :( i went as far as to message the dev on Quora, hopefully I'll get a response

bibinkt commented 5 years ago

Hey guys,

We can develop the wieght by our self . but its littil bit of manual effort are there , if you are interested we can work together and get it done !!

On Sat, Mar 16, 2019 at 10:05 AM Moe notifications@github.com wrote:

i am also looking for it :( i went as far as to message the dev on Quora, hopefully I'll get a response

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/sowmyadvn/Crime-Detection--using-Deep-learning/issues/3#issuecomment-473539672, or mute the thread https://github.com/notifications/unsubscribe-auth/AMaPotC0DOk9jQe3y84QevkhlWBFjXZzks5vXQhXgaJpZM4aYtWQ .

moeiscool commented 5 years ago

ive started developing one this morning based on this tutorial https://github.com/spmallick/learnopencv/tree/master/YOLOv3-Training-Snowman-Detector

read over all the steps and before you start modify this file https://github.com/spmallick/learnopencv/blob/master/YOLOv3-Training-Snowman-Detector/getDataFromOpenImages_snowman.py

change the classes to an array with your desired thingies. usable class names i found here https://storage.googleapis.com/openimages/2018_04/class-descriptions-boxable.csv

I'm doing this.

["Handgun","Rifle","Shotgun","Knife","Sword","Weapon","Screwdriver","Baseball bat","Flashlight","Dagger","Hammer","Torch","Person"]

im still in the downloading phase because "Person" has over 100,000 images 🤦‍♂️

I know this tutorial works because I followed it exactly as it was and made a snowman detector. So I expect it will do the same for my current selection of classes 🤞

bibinkt commented 5 years ago

Gr8 buddy .. you are awesome :-) let me try as well.. i was trying to pull the image from google and else where :-(

On Sat, Mar 16, 2019 at 9:12 PM Moe notifications@github.com wrote:

ive started developing one this morning based on this tutorial

https://github.com/spmallick/learnopencv/tree/master/YOLOv3-Training-Snowman-Detector

read over all the steps and before you start modify this file

https://github.com/spmallick/learnopencv/blob/master/YOLOv3-Training-Snowman-Detector/getDataFromOpenImages_snowman.py

change the classes to an array with your desired thingies. usable class names i found here

https://storage.googleapis.com/openimages/2018_04/class-descriptions-boxable.csv

["Handgun","Rifle","Shotgun","Knife","Sword","Weapon","Screwdriver","Baseball bat","Flashlight","Dagger","Hammer","Torch","Person"]

im still in the downloading phase because "Person" has over 100,000 images 🤦‍♂️

I know this tutorial works because I followed it exactly as it was and made a snowman detector. So I expect it will do the same for my current selection of classes 🤞

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/sowmyadvn/Crime-Detection--using-Deep-learning/issues/3#issuecomment-473609293, or mute the thread https://github.com/notifications/unsubscribe-auth/AMaPoinE5PGWbpd_pTYoMForOGT911aUks5vXaSWgaJpZM4aYtWQ .

myounus96 commented 4 years ago

Why you wanna train the model yourself when you get yolov3 model (check the first one) trained on openImages dataset .openImages have 'Person' class. @moeiscool And @bibinkt you can also use that because openImages also have classes related to weapons.

bibinkt commented 4 years ago

Yeah I tested that model but those are not effective but it works fine for person and other objects

Sent from my iPhone

On Oct 21, 2019, at 8:48 AM, Muhammad Younus notifications@github.com wrote:

 Why you wanna train the model yourself when you get yolov3 model (check the first one) trained on openImages dataset .openImages have 'Person' class. @moeiscool And @bibinkt you can also use that because openImages also have classes related to weapons.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or unsubscribe.

myounus96 commented 4 years ago

yup for some input videos it is unable to detect weapons,so did you came up with something better,can you please share?Thanks @bibinkt