6
votes
How does the bounding box regressor work in Fast R-CNN?
A very clear and in-depth explanation is provided by the slow R-CNN paper by Author(Girshick et. al) on page 12: C. Bounding-box regression and I simply paste here for quick reading:
Moreover, the ...
- 328
4
votes
How does the bounding box regressor work in Fast R-CNN?
The paper cited does not mention linear regression at all. What it does is using a neural network to predict continuous variables, and refers to that as regression.
The regression that is defined (...
- 6,111
4
votes
How does YOLO algorithm detect objects if the grid size is way smaller than the object in the test image?
Each grid predictor in YOLO should only have a high score that an object is within it, if it detects the centre of the bounding rectangle is inside itself. So a grid point that contains only the wing ...
- 29.1k
4
votes
what is darknet and why is it needed for YOLO object detection?
https://pjreddie.com/darknet/ is their website...
I cite :
"Darknet: Open Source Neural Networks in C
Darknet is an open source neural network framework written in C and CUDA. It is fast, easy to ...
- 2,378
4
votes
Accepted
What is the origin of YOLO/darknet coordinates
(0, 0) is top left.
Here an a helpful blog that goes through all the features in the output vector.
This is common in image processing. There are a few reasons ...
- 15k
4
votes
What are advantages or disadvantages of training deep learning model from scratch?
(Suggestions and edits will be appreciated)
let us discuss advantages of training a deep learning model from scratch:
Building and training NN from scratch is of a great use in the research field.
...
- 66
3
votes
What are advantages or disadvantages of training deep learning model from scratch?
The advantages of training a deep learning model from scratch and of transfer learning are subjective. It depends a lot on the problem you are trying to solve, the time constraints, the availability ...
- 31
3
votes
Accepted
How to label overlapping objects for deep learning model training
There is no common practice in labeling the bounding boxes. It is always problem dependent. For example, if you want to count the chickens then you should also label the whole chicken as one instance ...
- 1,938
3
votes
YOLO algorithm - understanding training data
For each bounding box you need
p_c: any object / no object (background)
b_x, b_y, ...
- 602
3
votes
Accepted
YOLO pretraining
1) The goal of using Average pooling layer (at least here), is to have a vector after it. That way you have a fully connected layer vector.
In Yolo, the layer previous the fully connected one seems ...
- 1,082
3
votes
Accepted
Last layers of YOLO
You can use the Flatten and Reshape layers to go to Dense and back to HWC format. The last layers in keras would look like this:
...
- 445
3
votes
Should there be a flat layer in between the conv layers and dense layer in YOLO?
Not sure that still matters for your project but it is important: the Dense layer does not flatten the entry first!
It takes the last dimension of the entry tensor and connects it to the neurons of ...
- 31
3
votes
Which is the "BEST" deep learning model for "Custom" object detection for images & real time. YOLO v3, v4, v5, EfficientDet?
There are many subtle differences between yolov4 and v5 other than speed like YOLOv4 exceeds YOLOv5's performance on the COCO benchmark.this link will help you explain.
YOLOv4 VS YOLOv5.
I dont know ...
- 707
3
votes
Car Make and Model detection
For detecting the make and model of cars from images with high accuracy across a large number of classes, I would recommend a convolutional neural network (CNN) architecture tailored for fine-grained ...
- 3,009
2
votes
Should there be a flat layer in between the conv layers and dense layer in YOLO?
It doesn't matter, with or without flattening, a Dense layer takes the whole previous layer as input. The spatial structure information is not used anymore. Some Neural Network implementations might ...
- 9,388
2
votes
what is darknet and why is it needed for YOLO object detection?
Darknet is mainly for Object Detection, and have different architecture, features than other deep learning frameworks. It is faster than many other NN architectures and approaches like FasterRCNN etc. ...
- 31
2
votes
Why does Pascal VOC 2007 dataset have almost 2500 images for 'train' and same number for 'val'? Val should have less images
The size of your validation matters only for the precision of your validation score. Every sample in your validation has some score for validation. The main goal is to be able to say how well your ...
- 9,388
2
votes
YOLO annotation guildelines: overlapping and partially visible objects
There is only 1 restriction, your truth bboxes with the same class_id shouldn't be overlapped more than IoU > 45%, because Yolo uses nms_threshold = 0.45.
In general, you should mark your objects in ...
- 649
2
votes
Does YOLO give preference to color over shape or vice-versa while detecting an object?
YOLO has 3 input channels (RGB typically) and is a CNN, a neural net that uses stacked (trainable) convolution filters. Single convolution filters act as feature detectors and can be (trained to be) ...
- 1,364
2
votes
how do the number of classes in an object detection model affect accuracy?
The general consensus in machine learning problems is that it becomes tougher to get higher accuracy results when there is more data with more class splits. The simplest of examples would be cifar 10 ...
- 482
2
votes
Accepted
Yolov3 Tiny: What do each of the 2535 cells detect?
As the blog mentioned, each cell predicts three things -
bbox coords (tx,ty,tw,th)
objectness score (po)
class scores (p0 - pc)
and again each cell predicts three boxes. Hence you get that big red ...
- 854
2
votes
Accepted
Training a YOLO-style object detector
As rightly said by @Nikos M., it is based on trial and error. And here are some tips you might find useful -
Create a good enough validation set.
Use YOLO-tiny versions instead of custom architecture....
- 854
2
votes
Computing F1 score for YOLOV5
The F-measure is the weighted harmonic mean of precision (P) and recall (R) of a classifier, taking α=1 (F1 score). It means that both metrics have the same importance. In your graph, the confidence ...
- 156
2
votes
Algorithms to do a CTRL+F (find object) on an image
Invariant object recognition(IOR), refers to rapid and accurate recognition of objects in the presence of variations such as size, rotation and position.
SIFT and SURF are the most popular among them, ...
- 1,259
2
votes
Should I remove objects labeled "unknown" from my test set?
It depends how you define the task, i.e. what is the goal of the model:
In standard classification, an "unknown" category doesn't make sense because there is no homogeneous "unknown&...
- 25.7k
2
votes
Accepted
How to interpret annotation data?
The format description can be found on file DATASET.md at the github repo you linked:
The bounding box (bbox) format is [top left X position, top left Y position, top left Z position, deltaX, deltaY, ...
- 27.2k
1
vote
YOLO Dense Prediction
It is kind of unfortunate that no further explanations were given in the paper about this. In my opinion Hard negative mining is actively used by architectures like SSD, actually by boosting its ...
- 11
1
vote
YOLO: How many bounding boxes?
From 98 boxes to 3 boxes, it involve many other things as well.
x*y*2 = 98, where 2 are the anchor boxes i.e. each grid will predict two bounding box.
Non Max Suppression: As correctly said ...
- 1,280
1
vote
How YOLO training and prediction works for an object fall in multiple grid?
It seems after referring many documents, I found the answer of my question.
First, likely to correct my understanding. I thought for labeling,
bounding box size (width, height) will always be ...
- 1,280
1
vote
what is darknet and why is it needed for YOLO object detection?
This deep learning framework is written itself in C but once you train the network you do not need Darknet itself for the inference. OpenCV has built in support for Darknet formats so both model and ...
Only top scored, non community-wiki answers of a minimum length are eligible