Hi My Name is Shahrukh Saleem i am a student of computer science and this is my final year project but i am facing a little bit problem first see my project working then i tell you what i want.
This project is designed to control railway level-crossing gate through the image processing techniques. In this system we are using machine learning algorithms to train our system so much that it can handle the frames/images extracted from the trains video, for this purpose first we have to collect the normal train arrival video how the train looks like when it is too far or how its look when its reach near to the camera, making a complete video of the train's arrival first the train is showing very small in the video and then it's become very large as its coming towards the camera and making number of data sets to be trained. So in this we used to capture the trains arrivals videos near the railway gates and extract the images from the videos and then we have to apply the machine learning algorithms on each frame extracted from the video to make train our system or make it so much efficient in this way the camera on the basis of machine learning the camera will detect the coming train toward it and then generates the signal towards the gates that the train is coming through the wire as the signal was sent the raspberry pi is attached with the gates or also connected with the camera through wires on receiving The signal the raspberry pie generates the alarming situations to alert the traffic or peoples on the railway gate crossings and a timer will start and gates shut down slowly timer is used because in case some peoples are inside the railway gate so they can immediately leave that area gates we used should not be so long enough because in some case when gates shut down or some traffic remains inside so the vehicle can go out through the corner as some space should be left to avoid any kind of mishap and gate area should be large not to close to the railway tracks because in some cases the vehicle will stick and cannot go out as there is much more traffic already there and there is no way to go so the space is left in case of that situation raspberry pi is connected with the DC motor to rotate the gate down so when trains leave the gate crossing area on behalf of no train in the video camera the gates again open as without train images are taken to make our system more efficient, so it would open the gates itself when there is no train showing in the camera. Our system is implemented on the prototype after the success of this system with more amendments we will later on implement this solution on real based environment.
the problem i facing is that i trained my model i have a little toy train engine i want to detect that train so on behalf of that train i will take decision the main problem is that the model i train or i see on the internet is good and working fine on laptop and PC but when i shift it to raspberry pi it lags too much and it cannot take decision due to this i am using the caffe pre_trained model mobile ssd_net so any one interested to do my project contact me i will provide data sets and please trained my data sets on better model which will easily run on the raspberry pi without any lag
8 freelancers are bidding on average $56 for this job
hi, bro. i am glad to meet you. I am a MATLAB, statistic and R expert.I have many experience in this field. I can help you 100% completely. Best regards!