PUBLICATIONS
Industri fashion selalu berkembang dengan pesat dimana ditandai dengan munculnya banyak merek baru, tipe atau desain setiap harinya. Salah satu produk fashion adalah produk tas. Tas bisa berupa tas tangan, tas ransel, tas pouch ataupun tas lainnya. Dengan selalu munculnya tipe baru pada produk tas, maka makin bertambah banyak jumlah tipe tas dari tiap merek yang ada. Kondisi yang demikian membuat masyarat agak susah untuk menghafal tipe tas tersebut dan di mana mereka bisa membeli tas tersebut. Tujuan dari penelitian ini adalah menghasilkan sebuah sistem pengenalan tas tangan menggunakan jaringan syaraf tiruan perambatan balik. Penelitian hanya terbatas pada pengenalan tipe tas wanita yang berupa tas tangan dan hanya pada tipe tas tertentu. Proses pengenalan tas tangan terdiri dari beberapa tahap yaitu tahap pengambilan gambar, tahap pre-processing, tahap ekstraksi fitur, tahap klasifikasi. Jaringan syaraf tiruan yang digunakan memiliki tiga lapisan yaitu lapisan masukan dimana terdiri dari 6 neuron, lapisan tersembunyi dimana terdiri dari 10 neuron dan lapisan keluran dimana terdiri dari 8 neuron. Pelatihan jaringan syaraf tiruan menggunakan algoritma peramabatan balik dengan nilai laju pelatihan sebesar 0,1 momentum sebesar 0,8. Pelatihan dilakukan menggunakan 414 data dengan nilai epoch sebesar 1.000 epoch dan nilai Mean Squared Error (MSE) sebesar 0.02. Pelatihan mencapai tingkat akurasi klasifikasi sebesar 71.1% dimana pelatihan berhenti pada epoch ke 980 dengan nila MSE sebesar 0.019641. Uji coba dilakukan menggunakan 28 data. Uji coba menunjukkan bahwa tingkat akurasi yang dihasilkan adalah 78,57%.
Implementation of Augmented Reality together with Cloud Service to improve flexibility and interactivity either for the end users as well as developers. Method/Statistical Analysis: Android has their precompiled OpenGL ES to draw graphics. Combining with OpenCV, it can result into a powerful enough Augmented Reality application and it is free to use. OpenCV provides feature detection and recognition and tracking for markers and OpenGL to draw 3D graphics. Findings: Implementing cloud service with smartphone application proves to be more efficient and flexible instead of general closed environment application. Adding markers and 3D data doesn't require the users to update their application each time, but it can automatically detect which marker are available in the server and display the appropriate 3D model depending on the marker detected. This provides the user with lower storage in their Smartphone's and better usage of the Augmented Reality application. Application/Implementation: This can be applied in all sorts of Virtual Reality or Augmented Reality application where the developers may add much data according to the users' needs.
Image background subtraction refers to the technique which eliminates the background in an image or video to extract the object (foreground) for further processing like object recognition in surveillance applications or object editing in film production, etc. The problem of image background subtraction becomes difficult if the background is cluttered, and therefore, it is still an open problem in computer vision. For a long time, conventional background extraction models have used only image-level information to model the background. However, recently, deep learning based methods have been successful in extracting styles from the image. Based on the use of deep neural styles, we propose an image background subtraction technique based on a style comparison. Furthermore, we show that the style comparison can be done pixel-wise, and also show that this can be applied to perform a spatially adaptive style transfer. As an example, we show that an automatic background elimination and background style transform can be achieved by the pixel-wise style comparison with minimal human interactions, i.e., by selecting only a small patch from the target and the style images.
In this paper, we propose a deep-image-prior-based demosaicing method for a random
RGBW color filter array (CFA). The color reconstruction from the random RGBW CFA is performed
by the deep image prior network, which uses only the RGBW CFA image as the training data. To
our knowledge, this work is a first attempt to reconstruct the color image with a neural network
using only a single RGBW CFA in the training. Due to the White pixels in the RGBW CFA, more
light is transmitted through the CFA than in the case with the conventional RGB CFA. As the image
sensor can detect more light, the signal-to-noise-ratio (SNR) increases and the proposed demosaicing
method can reconstruct the color image with a higher visual quality than other existing demosaicking
methods, especially in the presence of noise. We propose a loss function that can train the deep
image prior (DIP) network to reconstruct the colors from the White pixels as well as from the red,
green, and blue pixels in the RGBW CFA. Apart from using the DIP network, no additional complex
reconstruction algorithms are required for the demosaicing. The proposed demosaicing method
becomes useful in situations when the noise becomes a major problem, for example, in low light
conditions. Experimental results show the validity of the proposed method for joint demosaicing
and denoising.