Home

Linked List Cycle

141. Linked List Cycle Solutions Solution 1: Use set to memorize visited node. Extra space Solution 2: Make visited node pointed to itself. No extra space Break original structure

Read more

Reverse Linked List

206. Reverse Linked List Solutions Solution 1: Iterative def reverseList(self, head): """ :type head: ListNode :rtype: ListNode """ if head == None: return head next = head.next last = None while head!=None: tmp = head.next head.next = last last = head head = tmp return last Solution 2: Re...

Read more

Image Alignment

Image Alignment Template matching Slow, combinatory, global solution Multi-scale template matching Faster, combinatory, locally optimal Local refinement based on some initial guess Fastest, locally optimal Translation: $\begin{bmatrix}x+p_1 \\ y + p_2\end{bmatrix} = \begin{bmatrix}1 & 0 & ...

Read more

Motion Estimation

  Feature Matching Object Detection scoring Feature matching across pairs of images and not feature detection (e.g. cornerness score). A lower score is a better match, since we use distance measure for comparison. Higher -> more likely to be an object threshold Keep matches i...

Read more

Recognition

Feature Matching An object as a collection of bag of features deals well with occlusion scale invariant rotation invariant position of parts is not important Works well for object instances Not great for generic object categories Pros Simple Efficient algorithms Robust to deformations Cons No spatial reasoning Ba...

Read more

Image Descriptors

Image Descriptors Represent surrounding region mathematically Color histogram Mean dy/dx values Texton histograms Feature Descriptors should be Invariant: shouldn’t change even if image is transformed Discriminative: should be highly unique for each point Raw Image Patch: Intensity Not invariant to geometry and appearance cha...

Read more

Local Features

Local Features Interest point detection keypoint detection, feature detection Characteristics of Good Features Repeatable The same feature can be found in several images despite geometric and photometric transformations Distinct Captures recognizably different information that allows it to be distinguished ...

Read more

Customizing Authentication

Customizing Authentication Django maintains a list of “authentication backends” that it checks for authentication. When somebody calls django.contrib.auth.authenticate(), Django tries authenticating across all of its authentication backends with lazy evaluation. Therefore, the order of authentication backend matters. Django’s default authenti...

Read more